No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Machine Learning Models and Bankruptcy Prediction
Abstract
There has been intensive research from academics and practitioners regarding models for predicting bankruptcy and default events, for credit risk management. Seminal academic research has evaluated bankruptcy using traditional statistics techniques (e.g. discriminant analysis and logistic regression) and early artificial intelligence models (e.g. artificial neural networks). In this study, we test machine learning models (support vector machines, bagging, boosting, and random forest) to predict bankruptcy one year prior to the event, and compare their performance with results from discriminant analysis, logistic regression, and neural networks. We use data from 1985 to 2013 on North American firms, integrating information from the Salomon Center database and Compustat, analysing more than 10,000 firm-year observations. The key insight of the study is a substantial improvement in prediction accuracy using machine learning techniques especially when, in addition to the original Altman’s Z-score variables, we include six complementary financial indicators. Based on Carton and Hofer (2006), we use new variables, such as the operating margin, change in return-on-equity, change in price-to-book, and growth measures related to assets, sales, and number of employees, as predictive variables. Machine learning models show, on average, approximately 10% more accuracy in relation to traditional models. Comparing the best models, with all predictive variables, the machine learning technique related to random forest led to 87% accuracy, whereas logistic regression and linear discriminant analysis led to 69% and 50% accuracy, respectively, in the testing sample. We find that bagging, boosting, and random forest models outperform the others techniques, and that all prediction accuracy in the testing sample improves when the additional variables are included. Our research adds to the discussion of the continuing debate about superiority of computational methods over statistical techniques such as in Tsai, Hsu, and Yen (2014) and Yeh, Chi, and Lin (2014). In particular, for machine learning mechanisms, we do not find SVM to lead to higher accuracy rates than other models. This result contradicts outcomes from Danenas and Garsva (2015) and Cleofas-Sánchez, García, Marqués, and Sénchez (2016), but corroborates, for instance, Wang, Ma, and Yang (2014), Liang, Lu, Tsai, and Shih (2016), and Cano et al. (2017). Our study supports the applicability of the expert systems by practitioners as in Heo and Yang (2014), Kim, Kang, and Kim (2015) and Xiao, Xiao, and Wang (2016).
... In recent years, with more data available, the field has shifted towards machine learning algorithms, because they make fewer assumptions about data (Shin and Lee, 2002;Wang et al., 2014), can deal with larger sets of observations and can test and select dozens of features simultaneously (Wang et al., 2011;Tsai et al., 2014). These latest models have shown superior accuracy rates (Barboza et al., 2017). ...
... However, despite having different modeling frameworks, most of these studies use similar data: information from relatively big, audited and/or public companies (Matenda et al., 2022;Kuizinienė et al., 2022). This observation holds true for research in a multitude of countries, like the United States (Campbell et al., 2008;LoPucki and Doherty, 2015;Barboza et al., 2017), Korea (Kim et al., 2015), Taiwan (Tsai et al., 2014), Spain (Cleofas-Sánchez et al., 2016) and Indonesia (Kisman and Krisandi, 2019). ...
... In terms of financial variables, we started with Altman's five ratios (Altman, 1968), adjusted for privately held companies. Additionally, we incorporated an operating margin variable (Barboza et al., 2017), an asset tangibility variable (Gilson et al., 1990;Campello and Giambona, 2013), and a size variable (Ohlson, 1980;Hotchkiss, 1995;LoPucki and Doherty, 2015) to capture relevant aspects of the firms' financial profiles. To provide a comprehensive analysis, we examined these variables both statically, considering the ratios in the year prior to the filing, and dynamically, assessing the percentage change of each variable from one year to two years prior to the filing (Carton and Hofer, 2006;Barboza et al., 2017). ...
The corporate bankruptcy prediction literature has traditionally relied on data from public, audited companies. However, the vast majority of firms worldwide are privately-held and lack the same level of scrutiny over their financial statements. As a result, these businesses usually produce less accurate and transparent accounting reports. Our research problem is to address this gap: how stakeholders deal with these less reliable information? Using a novel dataset of 503 private firms that filed for reorganization in Brazil between 2007 and 2020, we found that financial ratios had a significantly lesser effect on explaining default and bankruptcy than what previous research suggested, due in part to the lower information content in the accounting statements within our database. Instead, lenders seem to focus on harder-to-conceal variables, such as collateralizable assets, as well as on institutional factors, like proxies of financial statement quality. There is also concerning evidence that specialized attorneys can "work the system" in favor of distressed companies regardless of their financial fundamentals. Additionally, we found that machine learning models outperformed traditional statistical ones in different sorts of metrics, corroborating the literature on the superior performance of non-linear approaches on datasets having synergistic causality among its features.
... Naturalmente, todos os estudos incidem sobre modelos de previsão. No entanto, alguns estudos não têm como principal objetivo criar modelos preditivos das falências, insolvências ou dificuldades financeiras, mas sim aplicar novos algoritmos para problemas de previsão (e.g., algoritmos de machine learning) para ver de que forma estes permitem melhorar a qualidade das previsões (e.g., [19] [9]). Outros estudos pretendem criar de raiz o melhor modelo preditivo de falências, insolvências ou dificuldades financeiras (e.g., [23] [25]). ...
... Já no que respeita aos períodos (temporais) da análise constata-se igualmente muita diversidade, variando estes entre três meses [3] e 29 anos [19]. Assim, muitos estudos consideram nos seus modelos possíveis impactos de crises económicas, como a de 2008 (e.g., [19] [17]). ...
... Já no que respeita aos períodos (temporais) da análise constata-se igualmente muita diversidade, variando estes entre três meses [3] e 29 anos [19]. Assim, muitos estudos consideram nos seus modelos possíveis impactos de crises económicas, como a de 2008 (e.g., [19] [17]). ...
... Accounting, Market, Macroeconomic AUC, Accuracy Barboza et al. (2017) Machine-learning models Accounting, Market AUC, Accuracy Tobback et al. (2017) Smoothed wvRN, Support vector machines Accounting, Relational AUC Zelenkov and Volodarskiy (2021) Logistic regression, Machine-learning models Accounting, Macroeconomic Geometric mean, AUC of bankruptcy prediction models. They find that decision trees are relatively more accurate than neural networks, but have more rule nodes than desired. ...
... Tobback et al. (2017) find that a combination of financial and relational data improves the performance of bankruptcy prediction models, as a company is more likely to fail when one of its related companies already filed for bankruptcy. Barboza et al. (2017) compare the performance of different bankruptcy prediction models based on the sensitivity, specificity and area under the curve, and find that machine-learning models outperform logistic regressions and linear discriminant analyses, as well as that random forests show the best performance among the machine-learning models. Our analysis builds on their approach and extends their work by introducing different costs of Type I and Type II errors. ...
... Since there is no unique set of variables used in previous prediction models, we consider three different sets of variables for our prediction models. These variable sets, based on accounting literature (Altman, 1968;Barboza et al., 2017;Ohlson, 1980) and shown in Table 2, are commonly used ratios for bankruptcy prediction models and are easily obtained from a company's balance sheet. Altman (1968) selected five ratios based on their statistical significance, evaluation of intercorrelations between the relevant variables, their predictive accuracy, and his judgement. ...
... With the development of machine learning and deep learning, plenty of studies have attempted to make a breakthrough by applying new models into bankruptcy prediction, and directly use well-calculated financial ratios to find the most predictive model and make predictions [4], [6], [9], [10], [22], [25]. In the early stage of applying machine learning methods for bankruptcy prediction, logistic regression [2] was once the most widely-used model in predicting bankruptcy. ...
... Even today, many financial institutions still adopt the logistic regression as the primary approach for building the credit scorecards because of its interpretability and stability [27] [23]. However, the results of using learning techniques have been shown to be contradictory at times, for instance, [22] and [4]. These discrepancies are due to the fact that machine learning models' results are highly dependent on the input data, whereas the financial data of firms, especially of SMEs, used to predict credit risk and bankruptcy are often unstructured and incomplete [18]. ...
... Hence, in this section, we limit ourselves to discussing some recent contributions that analyze the accuracy of ML and statistical classifiers. Barboza et al. (2017) conducts a comparative assessment of the bankruptcy prediction performance of support vector machines, bagging, boosting, ran-dom forests and neural networks with respect to some statistical models (discriminant analysis, logistic regression). The paper uses data on North American firms from 1985 to 2013, integrating information from the Salomon Center database and Compustat and analyzing more than 10 000 firm-year observations. ...
Default prediction is the primary goal of credit risk management.
This problem has long been tackled using well-established statistical classification models. Still, nowadays, the availability of large datasets and cheap software implementations makes it possible to employ machine learning techniques. This paper uses a large sample of small Italian companies to compare the performance of various machine learning classifiers and a more traditional logistic regression approach. In particular, we perform feature selection, use the algorithms for default prediction, evaluate their accuracy and find a more suitable threshold as a function of sensitivity and specificity. Our outcomes suggest that
machine learning is slightly better than logistic regression. However, the relatively small performance gain is insufficient to conclude that classical statistical classifiers should be abandoned, as they are characterized by more straightforward interpretation and implementation.
... Despite their widespread use in both academia and industry, these types of models have proven to be about numbers and quantities, necessitating the need for improvements that 78 span beyond numbers [19]. To address the constraint of these models, various research that employs pattern matching approaches has been substantially researched in the field of machine learning [20]. Several of which have proved machine learning models' ability to deal with unbalanced datasets [21], pictorial data and text data [22]. ...
This study examines machine learning methods used in crisis management. Analyzing detected patterns from a crisis involves the collection and evaluation of historical or near-realtime datasets through automated means. This paper utilized the meta-review method to analyze scientific literature that utilized machine learning techniques to evaluate human actions during crises. Selected studies were condensed into themes and emerging trends using a systematic literature evaluation of published works accessed from three scholarly databases. Results show that data from social media was prominent in the evaluated articles with 27% usage, followed by disaster management, health (COVID) and crisis informatics, amongst many other themes. Additionally, the supervised machine learning method, with an application of 69% across the board, was predominant. The classification technique stood out among other machine learning tasks with 41% usage. The algorithms that played major roles were the Support Vector Machine, Neural Networks, Naive Bayes, and Random Forest, with 23%, 16%, 15%, and 12% contributions, respectively.
... Several supervised machine learning classification method algorithms were chosen to get the best accuracy value [9], [10]. The algorithms are logistic regression (LR) [11], [12], naive bayes (NB) [13], [14], random forest (RF) [15], [16], k-nearest neighbor (KNN) [17], [18], and support vector machine (SVM) [19], [20]. Selected five classifications supervised machine learning based on each algorithm have on different dimension metrics there are parametric-simple for LR and NB, parametric-complex for SVM, non-parametric-simple for KNN, and non-parametic-complex for RF. ...
Machine breakdowns in the production line mostly finish in more than 18 minutes, since the machine that needs repair more time is done on the production line, not in the machine warehouse. Historical machine breakdown data is digitally recorded through the Andon system, but it is still not being used adequately to aid decision-making. This research introduces an analysis of historical machine breakdown data to provide predictions of repair time intervals with a focus on finding the best algorithm accuracy. The research method uses machine learning techniques with a classification model. There are five algorithms used: logistic regression (LR), naive bayes (NB), k-nearest neighbor (KNN), support vector machine (SVM), and random forest (RF). The results of this study prove that historical machine breakdown data can be optimized to predict machine repair time intervals in the production line. The accuracy of LR algorithm is slightly better than the other algorithms. Based on the receiver operating characteristic-area under curve (ROC-AUC) performance evaluation metric, the quality value of the accuracy of LR model is satisfied with a percentage of 69% with a difference of 0.5% between the train and test data.
... This analysis provides empirical results on users' intention to adopt financial technology during the COVID-19 pandemic. The results include an analysis of the model's performance based on several metrics, such as accuracy, precision, F1score, recall, and AUROC, consistent with previous analyses (Akour et al., 2021;Allen et al., 2022;Barboza et al., 2017;Y. Li & Li, 2020;Rodriguez-Galiano et al., 2015). ...
The novel coronavirus caused a lifestyle shift, and the acceptance of offsite financial transactions is still a case for financial technology (fintech). Mobile financial transactions continue to be at an all-time low, and financial institutions are developing approaches for financial digitalization acceptability. The present study attempts to understand users’ motivations for fintech adoption. The technology acceptance model (TAM) and the unified theory of acceptance and use of technology (UTUAT) were utilized to uncover the rationale behind technology adoption. This study explored the drivers inhibiting the adoption of financial technology in Nigeria during the pandemic. A machine learning (ML) approach was implemented to examine fintech adoption predictors using a self-administered consumer survey of 480 account holders. Survey responses were analyzed using a set of ML models (naïve Bayes, logistic regression, K-nearest neighbors, decision trees, and support vector machines), revealing the features and decision criteria for predicting perceived technology adoption. The decision tree outperformed the other models, with an accuracy of over 84%, precision of 88%, recall of 86%, F1-score of 84%, and area under the curve of 87%. The result indicates that customers are concerned about their safety. Thus, furthering their sense of risk. These results provide a roadmap for financial institutions and policymakers to understand behavioral attitudes toward adopting fintech and suggest strategies for attracting customers to the fintech space.
... This analysis provides empirical results on users' intention to adopt financial technology during the COVID-19 pandemic. The results include an analysis of the model's performance based on several metrics, such as accuracy, precision, F1score, recall, and AUROC, consistent with previous analyses (Akour et al., 2021;Allen et al., 2022;Barboza et al., 2017;Y. Li & Li, 2020;Rodriguez-Galiano et al., 2015). ...
The novel coronavirus caused a lifestyle shift, and the acceptance of offsite financial transactions is still a case for financial technology (fintech). Mobile financial transactions continue to be at an all-time low, and financial institutions are developing approaches for financial digitalization acceptability. The present study attempts to understand users' motivations for fintech adoption. The technology acceptance model (TAM) and the unified theory of acceptance and use of technology (UTUAT) were utilized to uncover the rationale behind technology adoption. This study explored the drivers inhibiting the adoption of financial technology in Nigeria during the pandemic. A machine learning (ML) approach was implemented to examine fintech adoption predictors using a self-administered consumer survey of 480 account holders. Survey responses were analyzed using a set of ML models (naïve Bayes, logistic regression, K-nearest neighbors, decision trees, and support vector machines), revealing the features and decision criteria for predicting perceived technology adoption. The decision tree outperformed the other models, with an accuracy of over 84%, precision of 88%, recall of 86%, F1-score of 84%, and area under the curve of 87%. The result indicates that customers are concerned about their safety. Thus, furthering their sense of risk. These results provide a roadmap for financial institutions and policymakers to understand behavioral attitudes toward adopting fintech and suggest strategies for attracting customers to the fintech space.
... In machine learning, data classification plays a very important role, aiming at extracting feature vectors from labeled dataset to build a causal relationship between labeled input and output pairs, then using the feature vectors to implement classification of unlabeled or new input data. Up to now, a large number of data classification methods have emerged and powered the development of machine learning and its practical applications in different domains (Jordan et al., 2015), such as image detection (Maggiori et al., 2017), speech recognition (Hinton et al., 2012), text understanding (Kamran et al., 2019), disease diagnosis (Bendi et al., 2011;Deepti et al., 2018), and financial prediction (Barboza et al., 2017). ...
Inspired by the ability of human ears to distinguish between different kinds of music and the ability of an ancient instrument to play countless beautiful melodies, this paper proposes the Rayleigh-quotient-based data classification method, which implements data classification by selecting an appropriate eigenfunction coordinate system to represent different categories of data and making different categories of data show significant differences in the coordinate components, meanwhile, transforming the selection problem of an eigenfunction coordinate system into a design problem of physical parameters of music instruments, such as designing a non-uniform string or membrane with distributed supports. In theory, the method is based on the eigenvalue problem for Sturm-Liouville differential equation, the eigenvalue problem for functional minimization, and the generalized eigenvalue problem for matrices, which have a time-honored history and can provide strong theoretical foundation and support. In addition, each parameter in the method has a clear physical interpretation and can be adjusted under the guidance of a set of complete theory, and their influence on classification results is definite, avoiding the large time consumption due to randomly adjusting parameters. The data classification procedure is divided into 2 steps, including mass density function design and stiffness function design, which not only simplifies the process of parameter adjustment, but also enables the optimization algorithm to be implemented by the linear programming with a low computational complexity, theoretically guaranteeing a high computational efficiency. Considering the amount of data being classified is large, two kinds of algorithms are developed to implement data classification, where one is a statistical averaging algorithm and the other is an incremental iterative algorithm. Finally, by combining the image database MNIST, the Rayleigh-quotient-based data classification method is employed to build an image classifier, which is trained on the MNIST training set including 60,000 images, and is tested on the MNIST test set including 10,000 images. The classification results show that by selecting an appropriate eigenfunction coordinate system, the Rayleigh quotient distribution of each category of image can be controlled within a very narrow range, while different categories of images show significant difference in the Rayleigh quotient distribution; the classification accuracy on the MNIST training set and on the MNIST test set are stable at about 99.50%, and the maximum accuracy can reach to 100.00% on the MNIST training set and 99.76% on the MNIST test set; and the Rayleigh-quotient threshold can be selected within a very wide range, beneficial for the robustness and stability of data classification.
... The basic goal is to assess the likelihood of companies' default by looking for relationships among different types of financial data, and the financial status of a firm in the future [30]. Barboza et al. [7] show that, on average, ML models exhibit 10% higher accuracy than scoring-based ones [55,77]. Specifically, Support Vector Machines (SVM), Random Forests (RF), as well as bagging and boosting techniques were tested for predicting bankruptcy events and compared with results from the discriminant analysis, Logistic Regression, and Neural Networks. ...
Modern financial markets produce massive datasets that need to be analysed using
new modelling techniques like those from (deep) Machine Learning and Artificial
Intelligence. The common goal of these techniques is to forecast the behaviour of the
market, which can be translated into various classification tasks, such as, for instance,
predicting the likelihood of companies’ bankruptcy or in fraud detection systems.
However, it is often the case that real-world financial data are unbalanced, meaning
that the classes’ distribution is not equally represented in such datasets. This gives the
main issue since any Machine Learning model is trained according to the majority
class mainly, leading to inaccurate predictions. In this paper, we explore different data
augmentation techniques to deal with very unbalanced financial data. We consider a
number of publicly available datasets, then apply state-of-the-art augmentation
strategies to them, and finally evaluate the results for several Machine Learning
models trained on the sampled data. The performance of the various approaches is
evaluated according to their accuracy, micro, and macro F1 score, and finally by
analyzing the precision and recall over the minority class. We show that a consistent
and accurate improvement is achieved when data augmentation is employed. The
obtained classification results look promising and indicate the efficiency of
augmentation strategies on financial tasks. On the basis of these results, we present
an approach focused on classification tasks within the financial domain that takes a
dataset as input, identifies what kind of augmentation technique to use, and then
applies an ensemble of all the augmentation techniques of the identified type to the
input dataset along with an ensemble of different methods to tackle the underlying
classification.
... In this regard, Cossin and Pirotte [18] suggest that the value should range between short-term debt and total debt, whereas Longstaff and Schwartz [19] suggest a methodology that considers a deterministic risk-free rate, a constant exogenous RR, and the possibility of bankruptcy occurring at any time with the inability to fulfil the performance obligation [20]. Finally, several studies [21,22] employ artificial intelligence and machine learning methods to select critical variables for bankruptcy prediction, whereas Boughaci et al. [23] propose a credit scoring model based on clustering techniques and random forests. ...
As the recent COVID-19 pandemic has clearly demonstrated, appropriate state support policies are crucial for supporting industrial supply chains during crises to prevent viable businesses from defaulting. In this context, this study proposes a hybrid credit risk model to appropriately size public interventions for viable (worthy) businesses through systematic risk assessment during a period of turmoil. This study discusses the effects of the credit crunch-based economic downturn and proposes a methodology to assist policymakers in managing limited public resources to effectively support industrial supply chains. The proposed approach initially focuses on the dynamics of credit risk during economic recession periods, identifying the conditions that may justify a public intervention strategy based on public guarantees. Subsequently, a hybrid credit risk model is developed to appropriately size public interventions by quantifying systematic risk. Finally, a numerical application is presented to demonstrate the effectiveness of the proposed approach.
... Typically, such a system contains a knowledge base embodying accumulated experience and a set of rules (inference engine) for applying the knowledge base to each particular situation that is described to the program. Expert systems of this type have been incorporated into financial failure models (Barboza et al., 2017;Metaxiotis & Psarras, 2003;Moynihan et al., 2006;Muñoz-Izquierdo et al., 2019aPavaloaia, 2009;Shin & Lee, 2002;Shiue et al., 2008;Ziba et al., 2016). ...
The objective of this paper is to critically evaluate the main weaknesses associated with the limitations of financial failure prediction research studies. For more than 80 years, researchers have unsuccessfully studied ways to create a general theory of financial failure, which is useful for prediction. In this paper, we review the main boundaries of failure prediction research through a critical evaluation of previous papers and our own approach from the research experience. Our findings corroborate that these studies suffer from a lack of theoretical and dynamic research, an unclear definition of failure, deficiencies with the quality of financial statement data and a shortfall in the diagnostic analyses of failure. The most relevant implications for future research in this area are also outlined. This is the first study to analyse in deep the caveats of financial failure prediction studies, a crucial topic nowadays due to the hints of an economic crisis caused by the Covid-19 pandemic.
El objetivo de este artículo es evaluar críticamente los principales puntos débiles asociados a las limitaciones de los estudios de investigación sobre predicción de quiebras financieras. Durante más de 80 años, los investigadores han estudiado sin éxito la forma de crear una teoría general del fracaso financiero que sea útil para la predicción. En este artículo, revisamos los principales límites de la investigación sobre predicción de quiebras mediante una evaluación crítica de trabajos anteriores y nuestro propio enfoque a partir de la experiencia investigadora. Nuestras conclusiones corroboran que estos estudios adolecen de una falta de investigación teórica y dinámica, una definición poco clara del fracaso, deficiencias con la calidad de los datos de los estados financieros y un déficit en los análisis de diagnóstico del fracaso. También se esbozan las implicaciones más relevantes para futuras investigaciones en este ámbito. Se trata del primer estudio que analiza en profundidad las salvedades de los estudios de predicción de la quiebra financiera, un tema crucial en la actualidad debido a los atisbos de crisis económica provocados por la pandemia del Covid-19.
... In addition, from the economic/financial perspective, data science models have been used in the financial sector to provide forecasts on stock market prices and bankruptcies (Barboza et al., 2017), to detect fraud activities (Pumsirirat & Yan, 2018), or to assess a client's creditworthiness by analysing information coming from several data sources. Credit default risk can be easily estimated based on AI models that account for different dimensions when banks give loans to private companies or individuals (Mancisidor et al., 2020). ...
Despite the abundance of studies focused on how higher education institutions (HEIs) are implementing sustainable development (SD) in their educational programmes , there is a paucity of interdisciplinary studies exploring the role of technology , such as data science, in an SD context. Further research is thus needed to identify how SD is being deployed in higher education (HE), generating positive exter-nalities for society and the environment. This study aims to address this research gap by exploring various ways in which data science may support university efforts towards SD. The methodology relied on a bibliometric analysis to understand and visualise the connections between data science and SD in HE, as well as reporting on selected case studies showing how data science may be deployed for creating SD impact in HE and in the community. The results from the bibliometric analysis unveil five research strands driving this field, and the case studies exemplify them. This study can be considered innovative since it follows previous research on artificial intelligence and SD. Moreover, the combination of bibliometric analysis and case studies provides an overview of trends, which may be useful to researchers and decision-makers who wish to explore the use of data science for SD in HEIs. Finally, the findings highlight how data science can be used in HEIs, combined with a framework developed to support further research into SD in HE.
... Several machine learning algorithms appeared to offer better performance than traditional statistical techniques. Barboza et al. have shown that Random Forest is superior to Altman's Z-score [16], which is a well-known financial ratio. Support Vector Machine is another effective algorithm for business failure, as well as credit scoring, which has yielded significant results in both [17] and [18]. ...
... Statistical analysis techniques are used to evaluate the difficulty of the challenges that are provided by machine learning technologies. In response to these challenges, Barboza et al. (2017) suggested the use of a "fuzzy support vector machine." This method of determining creditworthiness preserves the SVM's resistance to outliers by making use of algorithmic generalization, and it differentiates various creditors. ...
The management of modern enterprises is increasingly dependent on various forms of technological assistance. Applications of machine learning, artificial intelligence, and other types of algorithmic software have become some of the most pervasive impacts on commercial software. They provide a comprehensive range of services for the administration of businesses, one of which is support with the management of banking risks. The importance of risk management in the financial services industry, particularly banking and insurance, has expanded dramatically during the past decade. Throughout the course of their existence, financial institutions have traditionally placed a significant emphasis on risk monitoring, analysis, and reporting. But these days, in order to control risks in a more accurate and effective manner, they use machine learning. In light of this, the objective of this study was to conduct an investigation into the numerous applications of machine learning in banking risk management. In order to accomplish the objectives of the study, the researcher carried out a comprehensive literature review on the application of machine learning to banking risk management. According to the findings of the study, there is a plethora of academic and professional literature on the subject of transformations in the financial services industry, particularly as they pertain to risk management. It did a literature analysis on the subject, examined several different machine learning algorithms, and graded them according to their potential usefulness in risk management. It identified areas in which risk management may be enhanced and investigated potential solutions to the identified issues. The findings of the review indicate that there is a major information gap in the area of the potential role that machine learning could play in risk management within the financial services industry. Many research concentrated on credit risks, while liquidity, market, and operational risks were often disregarded in these investigations. On the other hand, it was discovered that applications of machine learning have the potential to be utilized to construct risk management models. Machine learning is applied to several data kinds for the purpose of conducting analysis in order to improve the accuracy with which likely occurrences are estimated and to compute losses associated with various sorts of risks. Furthermore, it was demonstrated that when it comes to the management of risks, techniques based on machine learning are superior to traditional statistical models in terms of their accuracy and precision. Even if machine learning makes it possible for banks to better manage risk, additional research is still required in a number of other areas.
... A variety of constructing algorithms used in Machine Learning allow computers to learn from data and make predictions and judgments based on that data. Machine learning's explosive growth over the last few decades has had a significant impact on how we live our daily lives with Examples of machine learning applications include predicting bankruptcy, predicting rainfall, forecasting the weather, self-driving systems, optical character recognition, and more [3] [4]. Even though machine ...
... Recently, except for the classical control methods mentioned above, machine learning-based methods have been used to improve the performance of mobile robot in target tracking by introducing novel methods in estimation of movement paths. Machine learning approach enables the robot to evaluate the received data based on artificial intelligence methods by computers [11]. Evolutionary computation and algorithms such as swarm intelligence algorithms are considered in trajectory tracking as a part of the machine learning family. ...
Mobile robots as automatic machines have the capability to move around various environments and have considered because of wide range of applications in the industrial settings. It is vital for a mobile robot to determine its errors in indeterminate conditions and provide a robust and high performance controller along with trajectory control. Hence, the research has attempted to give a learning-based controller method for wheeled mobile robots in the presence of disturbance. The proposed method, use LQR controller and the Bat algorithm based on machine learning agents to calculate and identify control errors and satisfy the stability. The wheeled mobile robot should track the reference path in the mode by observing and estimating errors. Compared to other presented methods, the control of wheeled mobile robots has been investigated in the presence of noise and unmodeled dynamic. The proposed algorithm is simulated in MATLAB software as virtual environment. The obtained results are compared with a classical method and is illustrated that the discussed method has reasonable performance compared to the classical method.
... ANNs are especially helpful for issues involving vast volumes of data and intricate interactions between variables. Although some studies preferred using boosting machines and random forests (Barboza et al., 2017;Sakri, 2022), ANNs were found by other studies to have similar or higher empirical predictive superiority (Ahmad et al., 2017;Begum, 2022;Naidu & Govinda, 2018). Knowing this, coupled with the widespread use of ANNs in the financial prediction literature, this research has opted to adopt ANNs for FDP modeling. ...
Aim/Purpose: This paper aims to empirically quantify the financial distress caused by the COVID-19 pandemic on companies listed on Amman Stock Exchange (ASE). The paper also aims to identify the most important predictors of financial distress pre- and mid-pandemic. Background: The COVID-19 pandemic has had a huge toll, not only on human lives but also on many businesses. This provided the impetus to assess the impact of the pandemic on the financial status of Jordanian companies. Methodology: The initial sample comprised 165 companies, which was cleansed and reduced to 84 companies as per data availability. Financial data pertaining to the 84 companies were collected over a two-year period, 2019 and 2020, to empirically quantify the impact of the pandemic on companies in the dataset. Two approaches were employed. The first approach involved using Multiple Discriminant Analysis (MDA) based on Altman’s (1968) model to obtain the Z-score of each company over the investigation period. The second approach involved developing models using Artificial Neural Networks (ANNs) with 15 standard financial ratios to find out the most important variables in predicting financial distress and create an accurate Financial Distress Prediction (FDP) model. Contribution: This research contributes by providing a better understanding of how financial distress predictors perform during dynamic and risky times. The research confirmed that in spite of the negative impact of COVID-19 on the financial health of companies, the main predictors of financial distress remained relatively steadfast. This indicates that standard financial distress predictors can be regarded as being impervious to extraneous financial and/or health calamities. Findings: Results using MDA indicated that more than 63% of companies in the dataset have a lower Z-score in 2020 when compared to 2019. There was also an 8% increase in distressed companies in 2020, and around 6% of companies came to be no longer healthy. As for the models built using ANNs, results show that the most important variable in predicting financial distress is the Return on Capital. The predictive accuracy for the 2019 and 2020 models measured using the area under the Receiver Operating Characteristic (ROC) graph was 87.5% and 97.6%, respectively. Recommendations for Practitioners: Decision makers and top management are encouraged to focus on the identified highly liquid ratios to make thoughtful decisions and initiate preemptive actions to avoid organizational failure. Recommendation for Researchers: This research can be considered a stepping stone to investigating the impact of COVID-19 on the financial status of companies. Researchers are recommended to replicate the methods used in this research across various business sectors to understand the financial dynamics of companies during uncertain times. Impact on Society: Stakeholders in Jordanian-listed companies should concentrate on the list of most important predictors of financial distress as presented in this study. Future Research: Future research may focus on expanding the scope of this study by including other geographical locations to check for the generalisability of the results. Future research may also include post-COVID-19 data to check for changes in results.
... They also discussed the managerial implications of using deep learning models, including the importance of data quality and the need for skilled data scientists. The authors of [2] explored the use of ML models for bankruptcy prediction. They compared the performance of traditional statistical models with that of various ML algorithms, including decision trees (DT), NNs, and support vector machines. ...
This paper presents a comparative analysis of traditional forecasting methods and machine learning (ML) techniques for sales prediction in e-commerce. We first review the literature on both traditional and ML methods for sales prediction in e-commerce, highlighting their strengths and weaknesses. The study uses a dataset of daily sales from an e-commerce retailer to conduct a comprehensive empirical study thar compares the performance of literature methods from both categories. The analysis considers different forecasting horizons and evaluates the accuracy of the predictions using various performance metrics, such as mean absolute error and mean squared error. The study finds that ML techniques generally outperform traditional methods, especially for longer forecasting horizons. However, some traditional methods, such as the Holt-Winters method, can also perform well under certain conditions. Our study provides insights into the relative strengths and weaknesses of traditional and ML methods for sales prediction in e-commerce and can guide practitioners in selecting appropriate methods for their specific requirements.
... Further scrutiny of the top-cited publications revealed the increasing usage of deep learning frameworks for financial time series and financial signal representation Deng et al. 2017). The research on credit scoring and bankruptcy prediction has also attracted significant scholarly attention (Barboza et al. 2017;Wang et al. 2011). Noteworthy articles involving the broader application of ML/DL mechanisms can also be found in research themes such as forecasting stock price index (Kim & Won 2018), statistical arbitrage (Krauss et al. 2017) and even business data mining (Bose & Mahapatra 2001). ...
This paper surveys the extant literature on machine learning, artificial intelligence, and deep learning mechanisms within the financial sphere using bibliometric methods. We considered the conceptual and social structure of publications in ML, AI, and DL in finance to better understand the research’s status, development, and growth. The study finds an upsurge in publication trends within this research arena, with a bit of concentration around the financial domain. The institutional contributions from USA and China constitute much of the literature on applying ML and AI in finance. Our analysis identifies emerging research themes, with the most futuristic being ESG scoring using ML and AI. However, we find there is a lack of empirical academic research with a critical appraisal of these algorithmic-based advanced automated financial technologies. There are severe pitfalls in the prediction process using ML and AI due to algorithmic biases, mostly in the areas of insurance, credit scoring and mortgages. Thus, this study indicates the next evolution of ML and DL archetypes in the economic sphere and the need for a strategic turnaround in academics regarding these forces of disruption and innovation that are shaping the future of finance.
... Specifically, we will employ the random forest algorithm proposed by [49,50], which was found to be the best model for short-term forecasting of the PD for crypto-assets with a long time series in [11]. Moreover, it has an excellent past track record in forecasting binary variables; see [22,[51][52][53] for more details. This algorithm aggregates multiple decision trees into a "forest", where each tree is constructed differently from the others to decrease the correlation among trees and prevent overfitting. ...
In this paper, we analyzed a dataset of over 2000 crypto-assets to assess their credit risk by computing their probability of death using the daily range. Unlike conventional low-frequency volatility models that only utilize close-to-close prices, the daily range incorporates all the information provided in traditional daily datasets, including the open-high-low-close (OHLC) prices for each asset. We evaluated the accuracy of the probability of death estimated with the daily range against various forecasting models, including credit scoring models, machine learning models, and time-series-based models. Our study considered different definitions of “dead coins” and various forecasting horizons. Our results indicate that credit scoring models and machine learning methods incorporating lagged trading volumes and online searches were the best models for short-term horizons up to 30 days. Conversely, time-series models using the daily range were more appropriate for longer term forecasts, up to one year. Additionally, our analysis revealed that the models using the daily range signaled, far in advance, the weakened credit position of the crypto derivatives trading platform FTX, which filed for Chapter 11 bankruptcy protection in the United States on 11 November 2022.
... Previous research on machine learning usually only predicts defaults through discrimination or calculation of credit scores rather than estimation of accumulative PDs for multiple periods. For example, Barboza et al. [25] predicted bankruptcy one year prior to the event and compared the performance of different machine learning models, including support vector machines, bagging, boosting, and random forest. Gunnarsson et al. [26] constructed a multilayer perceptron network and a deep belief network and compared their performance for credit scoring. ...
When predicting the defaults of a large number of samples in a region, this will be affected by industry default heterogeneity. To build a credit risk model that is more suitable for Chinese-listed firms, which have highly industry-specific default heterogeneity, we extend the forward intensity model to predict the defaults of Chinese-listed firms with information about the default heterogeneity of industries. Compared with the original model, we combine the Bayes approach with the forward intensity model to generate time-varying industry-specific default heterogeneity indicators. Our model can capture co-movements of different industries that cannot be observed based on the original forward intensity model so that the model can flexibly adjust the firm’s PD according to the industry. In addition, we also consider the impact of default heterogeneity in other industries by studying the influence of the level and trends of other industries’ default heterogeneity on a firm’s credit risk. Finally, we compute PDs for 4476 firms from January 2001 to December 2019 for 36 prediction horizons. The extended model improves the prediction accuracy ratios both for the in-sample and out-of-sample firm’s PDs for all 36 horizons. Almost all the accuracy ratios of the prediction horizons’ PDs are increased by more than 6%. In addition, our model also reduces the gap between the aggregated PDs and the realized number of defaults. Our industry-specific default heterogeneity indicator is helpful to improve the model’s performance, especially for predicting defaults in a large portfolio, which is of significance for credit risk management in China and other regions.
According to the literature review of technology acceptance designs, artificial intelligence is presented as a tool for technical and commercial use of crypto-currency, and it is based on several dimensions of multidimensional attractiveness hence the notion of blockchain. We are focused on an acceptance test according to the axes and technical orientations in order to know the tricks of changing the thoughts and behaviors of the human being. At this level, several theories have been linked (the New Technology Acceptance Model, Standardized framework for integrating new technologies, Theory of Predicted Behavior, using concepts from both the theory of flow and behavioral reasoning (TOF)) in a single model. To concretize our approach, we opt for a quantitative approach based on the PCA method (principal components analysis) in order to reduce the items and to answer the hypotheses posed at the beginning. We note from the results of our study the existence of a strong significance between the dimension “The technology tools of commercial intervention” and the variable to be explained “Technological efficiency” which is slightly significant which allows us to determine a deep acceptability of the blockchain systems.
İnsanoğlu makinelerin insanlar gibi düşünebildiği ve hareket edebildiği bir çağın başlangıcında bulunuyor. Bu durum her ne kadar ürkütücü görünse de, akademide ilgi gören ve üzerinde artan miktarda çalışmalar gerçekleşmeye başlanan bir konudur. Makine öğrenmesi ve derin öğrenmeyle oluşturulan yapay zeka pek çok alanda olduğu gibi, finans alanında da çokça kullanılmaktadır. Bu çalışmalar içerisinde yurt içinde ve yurt dışında yayınlanan makale, kitap, kitap bölümleri, sempozyum bildirileri olduğu gibi, gerçekleştirilen yüksek lisans ve doktora tezleri de yer almaktadır. Bu tür çalışmalarda gelinen son durumu tespit etmek ve literatürdeki boşlukları ortaya çıkarmak amacıyla sistematik literatür taramaları yapılmaktadır. Bu çalışmada, Türkiye’de gerçekleştirilen ve uygulama bölümlerinde finans alanında makine öğrenmesi ve derin öğrenme tekniklerinin kullanıldığı lisansüstü tezler sistematik literatür taraması ile incelenmektedir. Araştırma, 2018-2023 yılları arasındaki dönemde yapılan çalışmaları kapsamaktadır. Araştırmanın sonucunda, konuyla ilgili yapılan tezlerde makine ve derin öğrenme yöntemlerinin en çok finansal enstrümanların gelecekteki fiyatlarının tahminlemesinde, ardından sırasıyla finansal risklerin tespit edilmesinde, kurumsal finansal başarısızlık ve iflas tahmininde ve ayrıca portföy optimizasyon modellerinde kullanıldığı belirlenmiştir. İlgili dönem boyunca, yapılan tez çalışmalarının sayılarında artan bir grafik bulunmaktadır. Bu çalışmalarda genellikle birden fazla algoritmanın uygulamadaki başarıları karşılaştırılarak en başarılı sonuçlar belirlenmeye çalışılmıştır. En çok çalışılan tez konusunun makine öğrenmesiyle kredi riskinin analizi olduğu, ardından makine öğrenmesiyle hisse senedi fiyat tahmininin geldiği ortaya çıkmıştır. En çok kullanılan algoritmaların karar ormanı, karar ağacı ve uzun-kısa dönem hafıza algoritmaları olduğu tespit edilmiştir. Lisansüstü tez konusu olarak sosyal bilimlerden daha çok, fen bilimleri temel alanında tercih edildiği ve yazılan tezlerin en çok bilgisayar mühendisliği ana bilim dalında hazırlandığı, ardından işletme ana bilim dalında hazırlanan tezlerin geldiği ortaya çıkmıştır.
The financial literature regarding bankruptcy prediction models includes the application of several statistical tools which have progressively become accessible. This paper explores the impact of the Altman score on the stock prices of the Romanian companies listed on the Bucharest Stock Exchange during the period 2014-2022. The Altman score is calculated based on the annual reports of the analyzed companies collected from the Bloomberg database, while the stock exchange prices are collected from the Euromonitor database. The methodology used in the analysis is panel regression realized with the Generalized Method of Movements technique. The data is processed using the EViews software. This research study shows that there is a negative correlation between the Altman Z score and the companies’ share price. In terms of theoretical contribution, this study investigates one of the bankruptcy prediction models that is used more often than other models, namely the Altman Z-Score.
Early disease prediction can help sick persons determine the severity of the disease and take quick action, thus, a healthcare recommended system is viewed as additional tools to help patients control and manage their ill-health. Medical recommended system which provides users with quick and optimal disease predictions has been in existence for a while; however, it is faced with several data security issues. Sometimes, patients confidential data which are stored of the archive after each recommendation may be accessed by unauthorized persons, and this can warrant a serve data breach and disclosure of private medical information. Thus, the focus of our project is to design a privacy-aware recommended system that not just makes facilitates quick and easy recommendation for sick persons but also securely protects stored medical information from unauthorized access. This system will be designed to support quick search, recommendation septimal confidentiality and integrity.
Credit scoring models based on machine learning often need to work on accuracy and interpretability in practical applications. Original KCDWU has a more prominent adaptive property but ignores intra‐class and inter‐class distances in the clustering process, resulting in the possibility of inaccurate identification of class features and cluster structure of data, which compromises the clustering effect. Therefore, we improve the automatic K‐means clustering based on the Calinski–Harabasz index, thus achieving a clustering output for improved results. We also scrutinize representative five single classification models and six ensemble learning models for credit scoring prediction. We empirically test the superior performance of ensemble learning models and identify the best model CatBoost by comparing them based on multiple evaluation indicators. Empirical results reveal that the SHAP method conforms well to CatBoost and delivers a global and local interpretation of the predictions. This work provides financial institutions with a promising candidate for interpretable credit scoring models.
Recently, ensemble-based machine learning models have been widely used and have demonstrated their efficiency in bankruptcy prediction. However, these algorithms are black box models and people cannot understand why they make their forecasts. This explains why interpretability methods in machine learning attract attention from many artificial intelligence researchers. In this paper, we evaluate the prediction performance of Random Forest, LightGBM, XGBoost, and NGBoost (Natural Gradient Boosting for probabilistic prediction) for French firms from different industries with the horizon of 1–5 years. We then use Shapley Additive Explanations (SHAP), a model-agnostic method to explain XGBoost, one of the best models for our data. SHAP can show how each feature impacts the output from XGBoost. Furthermore, single prediction can also be explained, thus allowing black box models to be used in credit risk management.
Corporations getting bankrupt has been a severe issue for investors, businesses as well as ordinary individuals. Several research works have been conducted over the years to accurately predict bankruptcy, with the earliest works depending on the financial metrics of the company taken under consideration and the latest ones trying to predict bankruptcy from various kinds of data of organizations. However, the present works do not capture the dynamic nature of the business world and the possibility of a turnaround scenario. Hence, predictive models that are spatiotemporally aware when they predict a firm’s financial distress are needed. Considering this imminent problem, our work focuses on building a unique spatiotemporal context-aware bankruptcy prediction model that can predict bankruptcy with the help of daily news articles of a company or its related companies and key financial metrics to predict bankruptcy. Knowledge graphs were used to represent the vast amount of textual data. Their embeddings, along with the financial metrics, were used in the classification process. In the first stage, various machine learning algorithms were used for the financial metrics, while for the textual data or the embeddings, attention-based LSTM was used. Next, both were assembled together in the second stage to form the final predictive model, which has given an accuracy of 0.97 on the test set and an F1 score of 0.95. We hope our novel approach to this problem helps those who are uncertain about the future of any organization in predicting its bankruptcy beforehand and thereby timely decision making.
Corporate bankruptcy predictions are crucial to companies, investors, and authorities. However, most bankruptcy prediction studies have been based on stationary models, and they tend to ignore important challenges of financial distress like data non-stationarity, concept drift and data imbalance. This study proposes methods for dealing with these challenges and uses data collected from financial statements quarterly provided by companies to the Securities and Exchange Commission of Brazil (CVM). It is composed of information from 10 years (2011 to 2020), with 905 different corporations and 23,834 records with 82 indicators each. The sample majority have no financial difficulties, and only 651 companies have financial distress. The empirical experiment uses a sliding window, a history and a forgetting mechanism to avoid the degradation of the predictive model due to concept drift. The characteristics of the problem, especially the data imbalance, the performance of the models is measured through AUC, Gmean, and F1-Score and achieved 0.95, 0.68, and 0.58, respectively.KeywordsMachine LearningBankruptcyFinancial DistressData StreamData ImbalanceConcept DriftBrazilCVM
This chapter describes machine learning (ML) for financial applications with a focus on interpretable relational methods. It presents financial tasks, methodologies, and techniques in this ML area. It includes time dependence, data selection, forecast horizon, measures of success, quality of patterns, hypothesis evaluation, problem ID, method profile, and attribute-based and interpretable relational methodologies. The second part of this chapter presents ML models and practice in finance. It covers the use of ML in portfolio management, design of interpretable trading rules, and discovering money-laundering schemes using the machine learning methodology.KeywordsFinance time series Relational machine learning Visual knowledge discovery Decision tree Neural network Deep neural network Success measure Portfolio management Stock market Trading rules
The field of machine learning (ML) has grown to be a prominent subject within developed businesses while aiming to implement data-driven techniques to better day-to-day business activities. The reader is introduced to the most popular learning models in this chapter. Although unsupervised learning, reinforcement learning, and semi-supervised learning are incredibly important, the authors won't go into further detail about them here. This section will go into great length about supervised learning environments. The authors propose the following as a summary of the contributions to this chapter: Emphasizing some of the earlier works of literature that tackled these problems and discussing their limitations. In order to do so, the authors propose to review the regression family (i.e., simple regression and multiple linear regression) and decision tree family (i.e., CART, ID3, C4.5, chi-squared automatic interaction detection (CHAID), bagging and boosting). Examining the function and promise of ML techniques to resolve dilemmas and present potential implementation strategies.
Background and objective:
Exposure to solar ultraviolet (UV) radiation can cause malignant keratinocyte cancer and eye disease. Developing a user-friendly, portable, real-time solar UV alert system especially or wearable electronic mobile devices can help reduce the exposure to UV as a key measure for personal and occupational management of the UV risks. This research aims to design artificial intelligence-inspired early warning tool tailored for short-term forecasting of UV index (UVI) integrating satellite-derived and ground-based predictors for Australian hotspots receiving high UV exposures. The study further improves the trustworthiness of the newly designed tool using an explainable artificial intelligence approach.
Methods:
An enhanced joint hybrid explainable deep neural network model (called EJH-X-DNN) is constructed involving two phases of feature selection and hyperparameter tuning using Bayesian optimization. A comprehensive assessment of EJH-X- DNN is conducted with six other competing benchmarked models. The proposed model is explained locally and globally using robust model-agnostic explainable artificial intelligence frameworks such as Local Interpretable Model-Agnostic Explanations (LIME), Shapley additive explanations (SHAP), and permutation feature importance (PFI).
Results:
The newly proposed model outperformed all benchmarked models for forecasting hourly horizons UVI, with correlation coefficients of 0.900, 0.960, 0.897, and 0.913, respectively, for Darwin, Alice Springs, Townsville, and Emerald hotspots. According to the combined local and global explainable model outcomes, the site-based results indicate that antecedent lagged memory of UVI and solar zenith angle are influential features. Predictions made by EJH-X-DNN model are strongly influenced by factors such as ozone effect, cloud conditions, and precipitation.
Conclusion:
With its superiority and skillful interpretation, the UVI prediction system reaffirms its benefits for providing real-time UV alerts to mitigate risks of skin and eye health complications, reducing healthcare costs and contributing to outdoor exposure policy.
El objetivo de este artículo es presentar un análisis bibliométrico sobre el uso que han tenido las técnicas de Machine Learning (ML) en el proceso de predicción de quiebra empresarial a través de la revisión de la base de datos Web of Science. Este ejercicio brinda información sobre el inicio y el proceso de adaptación de dichas técnicas. Para ello, se identifican las diferentes técnicas de ml aplicadas en modelo de predicción de quiebras. Se obtiene como resultado 327 documentos, los cuales se clasifican por medida de evaluación del desempeño, área bajo la curva (AUC) y precisión (ACC), por ser las más utilizadas en el proceso de clasificación. Además, se identifica la relación entre investigadores, instituciones y países con mayor número de aplicaciones de este tipo. Los resultados evidencian que los algoritmos XGBoost, SVM, Smote, RFY DT presentan una capacidad predictiva mucho mayor que las metodologías tradicionales, enfocados en un horizonte de tiempo antes del suceso dada su mayor precisión. Así mismo, las variables financieras y no financieras contribuyen de manera favorable a dicha estimación.
Identifying and managing credit risk is vital for all lending institutions. Historically, credit risk is assessed using financial data from published financial statements. However, research indicates that the ability to detect financial hardship may be improved by textual analysis of firms’ disclosed records. This study aims to establish an association between themes and words from Management Discussion and Analysis (MDA) reports of firms and corporate failures. The study took a sample of 57 Indian listed firms declared bankrupt under the Insolvency and Bankruptcy Code (IBC) along with a matched sample of 55 solvent firms (matched by industry and size) for the period of FY2011–2019. The first part of analysis identifies negative words from the published reports and compares them with the negative words of the Loughran-McDonald dictionary. Then a thematic analysis is done to identify the key themes from the MDA reports and the significant themes are validated with their corresponding financial ratios in the third step using a panel logistic regression. Word analysis results show that IBC firms have significantly greater negative tone (2.21 percent) as against 1.30 percent of solvent firms. Thematic analysis results show that manageability, activity and performance are significant themes for predicting financial distress. Financial variables such as ownership pattern, promoters’ shares pledged, return on capital employed, asset utilization are some of the ratios in sync with the key themes. The study recommends that lenders and other stakeholders should look beyond financial statements which may be ‘window dressed’ by firms to qualitative disclosures in annual reports which may forewarn against impending financial distress. Acknowledgments The infrastructural support provided by FORE School of Management, New Delhi in completing this paper is gratefully acknowledged.
Gelişen teknolojiyle beraber diğer disiplinlerde olduğu gibi ormancılıkta da geleneksel uygulamaların daha ekonomik, etkin, hızlı ve kolay yapılabilmesi için yenilikçi yaklaşımların kullanımına talepler ve ihtiyaçlar artmaktadır. Özellikle son dönemde ortaya çıkan ormancılık bilişimi, hassas ormancılık, akıllı ormancılık, Ormancılık (Forestry) 4.0, iklim-akıllı ormancılık, sayısal ormancılık ve ormancılık büyük verisi gibi terimler ormancılık disiplinin gündeminde yer almaya başlamıştır. Bunların neticesinde de makine öğrenmesi ve son dönemde ortaya çıkan otomatik makine öğrenmesi (AutoML) gibi modern yaklaşımların ormancılıkta karar verme süreçlerine entegre edildiği akademik çalışmaların sayısında önemli artışlar gözlenmektedir. Bu çalışma, makine öğrenmesi algoritmalarının Türkçe dilinde anlaşılırlığını daha da artırmak, yaygınlaştırmak ve ilgilenen araştırmacılar için ormancılıkta kullanımına yönelik bir kaynak olarak değerlendirilmesi amacıyla ortaya konulmuştur. Böylece çeşitli ormancılık faaliyetlerinde makine öğrenmesinin hem geçmişten günümüze nasıl kullanıldığını hem de gelecekte kullanım potansiyelini ortaya koyan bir derleme makalesinin ulusal literatüre kazandırılması amaçlanmıştır.
The purpose of this work is to deepen the methodological provisions and develop practical recommendations with the purpose of increasing the competitiveness level of gas distribution network operators. To achieve this goal by systematizing and unifying different approaches, we developed our own approach to determining the competitiveness level of gas distribution network operators. Nowadays, most approaches and methods are formed with the obligatory use of expert assessment methods, which, in its turn, predetermines relatively subjective judgments and results. The results of our research have proven the significantly increased exigency for measures to prevent possible bankruptcy of gas distribution network operators functioning in the western region of Ukraine with the practical application of both foreign and domestic approaches. In the process of conducting a comprehensive analysis of financial and economic indicators and those reflecting the results of economic activity of gas distribution network operators functioning in the western region of Ukraine, the following approaches have been used in our study with the involvement of: 1) E. Altman’s Two-Factor Model; 2) E. Altman’s Five-Factor Model (Altman, 1977; 2020); 3) R. Lis’s bankruptcy prediction model; 4) Richard Taffler’s model; 5) W. Beaver’s coefficient; 6) O. Tereshchenko’s model; 7) Matviychuk’s model. Ultimately based on the study above, the following conclusions can be drawn as a method of assessing the of gas distribution network operators’ competitiveness: the existing models for diagnosing bankruptcy of enterprises are characterized by ambiguity; as for example, if R. Lis’s model indicates a low bankruptcy level, then other models prove the opposite situation; domestic diagnostic models need to be improved, as they were developed in the early 2000s and disregard current trends in functioning of enterprises. The article examines the methodological approaches to the level of competitiveness assessment of gas distribution network operators functioning in Ukraine. In our opinion, this fully corresponds to the thematic focus of the journal.
Researchers from all over the world have become more interested in bankruptcy prediction during the past 50 years. The prediction of corporate financial difficulties has been the subject of numerous studies ever since Altman revealed the breakthrough bankruptcy prediction model in 1968. The main aim of this research paper is to describe the fundamental concepts associated with the subject of corporate bankruptcy prediction. By identifying the most relevant research papers, nations, and authors in the Web of Science database, an in-depth review of the publications was performed before the analysis. The bibliometric map was created in the VOS Viewer program using the final search result with all available information. The results of the bibliometric analysis reveal that the keywords bankruptcy prediction and classification are the most closely related keywords using the analysis of citations that frequently occur, and that the USA and China developed the most significant international co-author relationships.
Tomatoes are used in almost every kitchen worldwide and are produced in large quantities. The productivity of tomato crops, both qualitatively and quantitatively, suffers from various infections. Besides, most tomato diseases commonly affect the leaves of the plant. The existing technique in tomato disease detection needs high human involvement and laborious tasks. Hence, this study designs and implements a convolutional neural network based on a robust autonomous tomato disease detection and classification system that fits the computational capabilities of resource-constrained devices. A custom-modified AlexNet architecture is used to detect diseases in tomato leaves. This approach relies on an automatic feature extraction process for classifying input images into respective disease classes. The model is trained and evaluated using Plant Village’s dataset, which contains approximately 14,000 diseased and healthy leaf photos separated into ten classifications. An average training classification accuracy of 99.67% and an average test classification accuracy of 98.43% were achieved in nine diseases and one healthy class by following different neural layer arrangements.KeywordsTomatoClassificationConvolutional Neural NetworkAlexNetPlant Village’s datasetautomatic feature extraction
Bankalar ve çeşitli finans kuruluşları tarafından karşılanan kredilerin, müşteri tarafından geri ödenememesi hem kredi veren kuruluşun sermaye kaybını hem de genel ekonomide oluşabilecek çeşitli risk faktörlerini beraberinde getirmektedir. Bu süreçte, oldukça kritik öneme sahip olan kredi riskinin doğru yönetilebilmesi ve uluslararası finans istikrarının sağlanması için Basel Komitesi ve BDDK (Bankacılık Düzenleme ve Denetleme Kurumu) gibi finans denetimi kuruluşları, kredi veren kurumların kredi verme karar aşamasında çeşitli regülasyon politikaları belirlemektedir. Ayrıca, kredi veren kurumlar analitik risk birimleri aracılığıyla kredi değerlendirme modelleri geliştirerek, müşterilere ait kredi risk skorunu hesaplamaktadır. Bu çalışmada makine öğrenmesi yöntemiyle kredi skorlama sistemlerinde kullanılabilecek en başarılı tahmini gerçekleştiren algoritmanın belirlenmesi amaçlanmıştır. Bu kapsamda, Gradyan Artırma, Yapay Sinir Ağları, Lojistik Regresyon, Rassal Orman, Karar Ağacı, Destek Vektör Makineleri, K-En Yakın Komşu ve WOE dönüşümleriyle Lojistik Regresyon algoritmaları için modeller kurulmuş ve temerrüde düşen ve temerrüde düşmeyen müşteriler için en iyi sınıflandırma performansı gösteren Gradyan Artırma algoritması olmuştur.
The author tested the hypothesis that default prediction based on financial data may be inapplicable to Russian service sector organizations by analyzing the differences in the accuracy of models based solely on financial data for service providers from Russia and developed European countries. Logistic regression, Random Forest and K-nearest neighbors machine learning methods were used as modeling tools on a sample of 404 Russian firms and 304 firms from developed European countries.The results suggest that the prediction error is significantly higher in the case of Russian firms than in the case of firms from the control group (European service firms). Thus, the use of financial ratios for default prediction for service firms in Russia seems insufficient. These findings can be used by organizations that provide credit scoring, and by any other market participants interested in the financial stability assessment of their counterparties.
The purpose of this paper is to examine the capital structure decisions of restaurant firms. The paper hypothesizes that these decisions are based upon a financial “pecking order” as well as the position of the firm in the financial growth cycle. Using ratios from publicly traded restaurant firms in the U.S. and ordinary least squares regression models, the results tend to support the notion that both the pecking order and the financial growth cycle influence financing decisions. However, the results also indicate that there may be separate factors affecting long-term and short-term debt decisions made by restaurant managers.
Technical and quantitative analysis in financial trading use mathematical and statistical tools to help investors decide on the optimum moment to initiate and close orders. While these traditional approaches have served their purpose to some extent, new techniques arising from the field of computational intelligence such as machine learning and data mining have emerged to analyse financial information. While the main financial engineering research has focused on complex computational models such as Neural Networks and Support Vector Machines, there are also simpler models that have demonstrated their usefulness in applications other than financial trading, and are worth considering to determine their advantages and inherent limitations when used as trading analysis tools. This paper analyses the role of simple machine learning models to achieve profitable trading through a series of trading simulations in the FOREX market. It assesses the performance of the models and how particular setups of the models produce systematic and consistent predictions for profitable trading. Due to the inherent complexities of financial time series the role of attribute selection, periodic retraining and training set size are discussed in order to obtain a combination of those parameters not only capable of generating positive cumulative returns for each one of the machine learning models but also to demonstrate how simple algorithms traditionally precluded from financial forecasting for trading applications presents similar performances as their more complex counterparts. The paper discusses how a combination of attributes in addition to technical indicators that has been used as inputs of the machine learning-based predictors such as price related features, seasonality features and lagged values used in classical time series analysis are used to enhance the classification capabilities that impacts directly into the final profitability.
Multi-layer perception (MLP) neural networks are widely used in automatic credit scoring systems with high accuracy and efficiency. This paper presents a higher accuracy credit scoring model based on MLP neural networks that have been trained with the back propagation algorithm. Our work focuses on enhancing credit scoring models in three aspects: (i) to optimise the data distribution in datasets using a new method called Average Random Choosing; (ii) to compare effects of training–validation–test instance numbers; and (iii) to find the most suitable number of hidden units. We trained 34 models 20 times with different initial weights and training instances. Each model has 6 to 39 hidden units with one hidden layer. Using the well-known German credit dataset we provide test results and a comparison between models, and we get a model with a classification accuracy of 87%, which is higher by 5% than the best result reported in the relevant literature of recent years. We have also proved that our optimisation of dataset structure can increase a model’s accuracy significantly in comparison with traditional methods. Finally, we summarise the tendency of scoring accuracy of models when the number of hidden units increases. The results of this work can be applied not only to credit scoring, but also to other MLP neural network applications, especially when the distribution of instances in a dataset is imbalanced.
Due to the economic significance of bankruptcy prediction of companies for financial institutions, investors and governments, many quantitative methods have been used to develop effective prediction models. Support vector machine SVM, a powerful classification method, has been used for this task; however, the performance of SVM is sensitive to model form, parameter setting and features selection. In this study, a new approach based on direct search and features ranking technology is proposed to optimise features selection and parameter setting for 1-norm and least-squares SVM models for bankruptcy prediction. This approach is also compared to the SVM models with parameter optimisation and features selection by the popular genetic algorithm technique. The experimental results on a data set with 2010 instances show that the proposed models are good alternatives for bankruptcy prediction.
This study aims to explore the relations between bank credit risks and macroeconomic factors. We employ a set of variables including the inflation rate, interest rate, the ISE-100 index, foreign exchange rate, growth rate, M2 money supply, unemployment rate, and the credit risk represented by the ratio of non-performing loans to total loans (NPL) for Turkey during the January 1998 and July 2012 period. The general-to-specific modelling methodology developed by Hendry (1980) was employed to analyze short-run dynamic intervariable relationships, while Engle-Granger (1987) and Gregory-Hansen (1996) methodologies were used to analyze long-run relationships. In both methods, growth rate and ISE index are the variables that reduce banks' credit risk in the long run, while money supply, foreign exchange rate, unemployment rate, inflation rate, and interest rate are the variables that increase banks' credit risks. The specific model demonstrated that the previous period's credit risk has a significant impact on the current period's credit risk.
Banking systemic risk is a complex nonlinear phenomenon and has shed light on the importance of safeguarding financial stability by recent financial crisis. According to the complex nonlinear characteristics of banking systemic risk, in this paper
we apply support vector machine (SVM) to the prediction of banking systemic risk in an attempt to suggest a new model with better explanatory power and stability. We conduct a case study of an SVM-based prediction model for Chinese banking systemic risk and find the experiment results showing that support vector machine is an efficient method in such case.
Support vector machines (SVMs), with their roots in Statistical Learning Theory (SLT) and optimization methods, have become powerful tools for problem solution in machine learning. SVMs reduce most machine learning problems to optimization problems and optimization lies at the heart of SVMs. Lots of SVM algorithms involve solving not only convex problems, such as linear programming, quadratic programming, second order cone programming, semi-definite programming, but also non-convex and more general optimization problems, such as integer programming, semi-infinite programming, bi-level programming and so on. The purpose of this paper is to understand SVM from the optimization point of view, review several representative optimization models in SVMs, their applications in economics, in order to promote the research interests in both optimization-based SVMs theory and economics applications. This paper starts with summarizing and explaining the nature of SVMs. It then proceeds to discuss optimization models for SVM following three major themes. First, least squares SVM, twin SVM, AUC Maximizing SVM, and fuzzy SVM are discussed for standard problems. Second, support vector ordinal machine, semisupervised SVM, Universum SVM, robust SVM, knowledge based SVM and multi-instance SVM are then presented for nonstandard problems. Third, we explore other important issues such as lp-norm SVM for feature selection, LOOSVM based on minimizing LOO error bound, probabilistic outputs for SVM, and rule extraction from SVM. At last, several applications of SVMs to financial forecasting, bankruptcy prediction, credit risk analysis are introduced.
Examines the capital structure decisions of restaurant firms. Hypothesizes that these decisions are based upon a financial “pecking-order” as well as the position of the firm in the financial growth cycle. Using ratios from publicly-traded restaurant firms in the USA and ordinary least squares regression models, the results tend to support the notion that both the pecking-order and the financial growth cycle influence financing decisions. However, the results also indicate that there may be separate factors affecting long-term and short-term debt decisions made by restaurant managers.
Empirical accounting researchers often use Altman's (1968) and Ohlson's (1980) bankruptcy prediction models as indicators of financial distress. While these models performed relatively well when they were estimated, we show that they do not perform as well in more recent periods (in particular, the 1980s), even when the coefficients are re-estimated. When we compare the performance of Ohlson's original model to our re-estimated version of his model and to that of Altman's original and re-estimated models, we find that Ohlson's original model displays the strongest overall performance. Given that Ohlson's original model is frequently used in academic research as an indicator of financial distress, its strong performance in this study supports its use as a preferred model.
In this paper, we investigate the performance of several systems based on ensemble of classifiers for bankruptcy prediction and credit scoring.The obtained results are very encouraging, our results improved the performance obtained using the stand-alone classifiers. We show that the method “Random Subspace” outperforms the other ensemble methods tested in this paper. Moreover, the best stand-alone method is the multi-layer perceptron neural net, while the best method tested in this work is the Random Subspace of Levenberg–Marquardt neural net.In this work, three financial datasets are chosen for the experiments: Australian credit, German credit, and Japanese credit.
The twenty-first century has seen a breathtaking expansion of statistical methodology, both in scope and in influence. ‘Big data’, ‘data science’, and ‘machine learning’ have become familiar terms in the news, as statistical methods are brought to bear upon the enormous data sets of modern science and commerce. How did we get here? And where are we going? This book takes us on an exhilarating journey through the revolution in data analysis following the introduction of electronic computation in the 1950s. Beginning with classical inferential theories - Bayesian, frequentist, Fisherian - individual chapters take up a series of influential topics: survival analysis, logistic regression, empirical Bayes, the jackknife and bootstrap, random forests, neural networks, Markov chain Monte Carlo, inference after model selection, and dozens more. The distinctly modern approach integrates methodology and algorithms with statistical inference. The book ends with speculation on the future direction of statistics and data science.
Understanding if credit risk is driven mostly by idiosyncratic firm characteristics or by systematic factors is an important issue for the assessment of financial stability. By exploring the links between credit risk and macroeconomic developments, we observe that in periods of economic growth there may be some tendency towards excessive risk-taking. Using an extensive dataset with detailed information for more than 30Â 000 firms, we show that default probabilities are influenced by several firm-specific characteristics. When time-effect controls or macroeconomic variables are also taken into account, the results improve substantially. Hence, though the firms' financial situation has a central role in explaining default probabilities, macroeconomic conditions are also very important when assessing default probabilities over time.
Identifying students’ learning styles has several benefits such as making students aware of their strengths and weaknesses when it comes to learning and the possibility to personalize their learning environment to their learning styles. While there exist learning style questionnaires for identifying a student's learning style, such questionnaires have several disadvantages and therefore, research has been conducted on automatically identifying learning styles from students’ behavior in a learning environment. Current approaches to automatically identify learning styles have an average precision between 66% and 77%, which shows the need for improvements in order to use such automatic approaches reliably in learning environments. In this paper, four computational intelligence algorithms (artificial neural network, genetic algorithm, ant colony system and particle swarm optimization) have been investigated with respect to their potential to improve the precision of automatic learning style identification. Each algorithm was evaluated with data from 75 students. The artificial neural network shows the most promising results with an average precision of 80.7%, followed by particle swarm optimization with an average precision of 79.1%. Improving the precision of automatic learning style identification allows more students to benefit from more accurate information about their learning styles as well as more accurate personalization towards accommodating their learning styles in a learning environment. Furthermore, teachers can have a better understanding of their students and be able to provide more appropriate interventions.
Thesupport-vector network is a new learning machine for two-group classification problems. The machine conceptually implements the following idea: input vectors are non-linearly mapped to a very high-dimension feature space. In this feature space a linear decision surface is constructed. Special properties of the decision surface ensures high generalization ability of the learning machine. The idea behind the support-vector network was previously implemented for the restricted case where the training data can be separated without errors. We here extend this result to non-separable training data.High generalization ability of support-vector networks utilizing polynomial input transformations is demonstrated. We also compare the performance of the support-vector network to various classical learning algorithms that all took part in a benchmark study of Optical Character Recognition.
The optimal selection of chemical features (molecular descriptors) is an essential pre-processing step for the efficient application of computational intelligence techniques in virtual screening for identification of bioactive molecules in drug discovery. The selection of molecular descriptors has key influence in the accuracy of affinity prediction. In order to improve this prediction, we examined a Random Forest (RF)-based approach to automatically select molecular descriptors of training data for ligands of kinases, nuclear hormone receptors, and other enzymes. The reduction of features to use during prediction dramatically reduces the computing time over existing approaches and consequently permits the exploration of much larger sets of experimental data. To test the validity of the method, we compared the results of our approach with the ones obtained using manual feature selection in our previous study (Perez-Sanchez, Cano, and Garcia-Rodriguez, 2014).The main novelty of this work in the field of drug discovery is the use of RF in two different ways: feature ranking and dimensionality reduction, and classification using the automatically selected feature subset. Our RF-based method outperforms classification results provided by Support Vector Machine (SVM) and Neural Networks (NN) approaches.
The task of classifying is natural to humans, but there are situations in which a person is not best suited to perform this function, which creates the need for automatic methods of classification. Traditional methods, such as logistic regression, are commonly used in this type of situation, but they lack robustness and accuracy. These methods do not not work very well when the data or when there is noise in the data, situations that are common in expert and intelligent systems. Due to the importance and the increasing complexity of problems of this type, there is a need for methods that provide greater accuracy and interpretability of the results. Among these methods, is Boosting, which operates sequentially by applying a classification algorithm to reweighted versions of the training data set. It was recently shown that Boosting may also be viewed as a method for functional estimation. The purpose of the present study was to compare the logistic regressions estimated by the maximum likelihood model (LRMML) and the logistic regression model estimated using the Boosting algorithm, specifically the Binomial Boosting algorithm (LRMBB), and to select the model with the better fit and discrimination capacity in the situation of presence(absence) of a given property (in this case, binary classification). To illustrate this situation, the example used was to classify the presence (absence) of coronary heart disease (CHD) as a function of various biological variables collected from patients. It is shown in the simulations results based on the strength of the indications that the LRMBB model is more appropriate than the LRMML model for the adjustment of data sets with several covariables and noisy data. The following sections report lower values of the information criteria AIC and BIC for the LRMBB model and that the Hosmer-Lemeshow test exhibits no evidence of a bad fit for the LRMBB model. The LRMBB model also presented a higher AUC, sensitivity, specificity and accuracy and lower values âÇïof false positives rates and false negatives rates, making it a model with better discrimination power compared to the LRMML model. Based on these results, the logistic model adjusted via the Binomial Boosting algorithm (LRMBB model) is better suited to describe the problem of binary response, because it provides more accurate information regarding the problem considered.
We assess whether two popular accounting-based measures, Altman’s (1968) Z-Score and Ohlson’s (1980) O-Score, effectively summarize publicly-available information about the probability of bankruptcy. We compare the relative information content of these Scores to a market-based measure of the probability of bankruptcy that we develop based on the Black–Scholes–Merton option-pricing model, BSM-Prob. Our tests show that BSM-Prob provides significantly more information than either of the two accounting-based measures. This finding is robust to various modifications of Z-Score and O-Score, including updating the coefficients, making industry adjustments, and decomposing them into their lagged levels and changes. We recommend that researchers use BSM-Prob instead of Z-Score and O-Score in their studies and provide the SAS code to calculate BSM-Prob.
This paper presents an alternative technique for financial distress prediction systems. The method is based on a type of neural network, which is called hybrid associative memory with translation. While many different neural network architectures have successfully been used to predict credit risk and corporate failure, the power of associative memories for financial decision-making has not been explored in any depth as yet. The performance of the hybrid associative memory with translation is compared to four traditional neural networks, a support vector machine and a logistic regression model in terms of their prediction capabilities. The experimental results over nine real-life data sets show that the associative memory here proposed constitutes an appropriate solution for bankruptcy and credit risk prediction,performing significantly better than the rest of models under class imbalance and data overlapping conditions in terms of the true positive rate and the geometric mean of true positive and true negative rates.
Ensemble techniques such as bagging or boosting, which are based on combinations of classifiers, make it possible to design models that are often more accurate than those that are made up of a unique prediction rule. However, the performance of an ensemble solely relies on the diversity of its different components and, ultimately, on the algorithm that is used to create this diversity. It means that such models, when they are designed to forecast corporate bankruptcy, do not incorporate or use any explicit knowledge about this phenomenon that might supplement or enrich the information they are likely to capture. This is the reason why we propose a method that is precisely based on some knowledge that governs bankruptcy, using the concept of “financial profiles”, and we show how the complementarity between this technique and ensemble techniques can improve forecasts.
Credit scoring aims to assess the risk associated with lending to individual consumers. Recently, ensemble classification methodology has become popular in this field. However, most researches utilize random sampling to generate training subsets for constructing the base classifiers. Therefore, their diversity is not guaranteed, which may lead to a degradation of overall classification performance. In this paper, we propose an ensemble classification approach based on supervised clustering for credit scoring. In the proposed approach, supervised clustering is employed to partition the data samples of each class into a number of clusters. Clusters from different classes are then pairwise combined to form a number of training subsets. In each training subset, a specific base classifier is constructed. For a sample whose class label needs to be predicted, the outputs of these base classifiers are combined by weighted voting. The weight associated with a base classifier is determined by its classification performance in the neighborhood of the sample. In the experimental study, two benchmark credit data sets are adopted for performance evaluation, and an industrial case study is conducted. The results show that compared to other ensemble classification methods, the proposed approach is able to generate base classifiers with higher diversity and local accuracy, and improve the accuracy of credit scoring.
Focusing on credit risk modelling, this paper introduces a novel approach for ensemble modelling based on a normative linear pooling. Models are first classified as dominant and competitive, and the pooling is run using the competitive models only. Numerical experiments based on parametric (logit, Bayesian model averaging) and nonparametric (classification tree, random forest, bagging, boosting) model comparison shows that the proposed ensemble performs better than alternative approaches, in particular when different modelling cultures are mixed together (logit and classification tree). Copyright
Effective bankruptcy prediction is critical for financial institutions to make appropriate lending decisions. In general, the input variables (or features), such as financial ratios, and prediction techniques, such as statistical and machine learning techniques, are the two most important factors affecting the prediction performance. While many related works have proposed novel prediction techniques, very few have analyzed the discriminatory power of the features related to bankruptcy prediction. In the literature, in addition to financial ratios (FRs), corporate governance indicators (CGIs) have been found to be another important type of input variable. However, the prediction performance obtained by combining CGIs and FRs has not been fully examined. Only some selected CGIs and FRs have been used in related studies and the chosen features may differ from study to study. Therefore, the aim of this paper is to assess the prediction performance obtained by combining seven different categories of FRs and five different categories of CGIs. The experimental results, based on a real-world dataset from Taiwan, show that the FR categories of solvency and profitability and the CGI categories of board structure and ownership structure are the most important features in bankruptcy prediction. Specifically, the best prediction model performance is obtained with a combination in terms of prediction accuracy, Type I/II errors, ROC curve, and misclassification cost. However, these findings may not be applicable in some markets where the definition of distressed companies is unclear and the characteristics of corporate governance indicators are not obvious, such as in the Chinese market.
Business health prediction is critical and challenging in today’s volatile environment, thus demand going beyond classical business failure studies underpinned by rigidities, like paired sampling, a-priori predictors, rigid binary categorization, amongst others.
In response, our paper proposes an investor-facing dynamic model for characterizing business health by using a mixed set of techniques, combining both classical and “expert system” methods. Data for constructing the model was obtained from 198 multinational manufacturing and service firms spread over 26 industrial sectors, through Wharton database.
The novel 4-stage methodology developed combines a powerful stagewise regression for dynamic predictor selection, a linear regression for modelling expert ratings of firms’ stock value, an SVM model developed from unmatched sample of firms, and finally an SVM-probability model for continuous classification of business health. This hybrid methodology reports comparably higher classification and prediction accuracies (over 0.96 and ~90%, respectively) and predictor extraction rate (~96%). It can also objectively identify and constitute new unsought variables to explain and predict behaviour of business subjects.
Among other results, such a volatile model build upon a stable methodology can influence business practitioners in a number of ways to monitor and improve financial health. Future research can concentrate on adding a time-variable to the financial model along with more sector-specificity.
There is great discussion but little consensus on the best measures of organizational performance. This book redresses this imbalance. Measuring Organizational Performance offers a framework with which to better understand the implications of selecting variables for use in both empirical studies and practice where organizational financial performance is the critical issue. © Robert B. Carton and Charles W. Hofer 2006. All rights reserved.
In parallel to the increase in the number of credit card transactions, the financial losses due to fraud have also increased. Thus, the popularity of credit card fraud detection has been increased both for academicians and banks. Many supervised learning methods were introduced in credit card fraud literature some of which bears quite complex algorithms. As compared to complex algorithms which somehow over-fit the dataset they are built on, one can expect simpler algorithms may show a more robust performance on a range of datasets. Although, linear discriminant functions are less complex classifiers and can work on high-dimensional problems like credit card fraud detection, they did not receive considerable attention so far. This study investigates a linear discriminant, called Fisher Discriminant Function for the first time in credit card fraud detection problem. On the other hand, in this and some other domains, cost of false negatives is very higher than false positives and is different for each transaction. Thus, it is necessary to develop classification methods which are biased toward the most important instances. To cope for this, a Modified Fisher Discriminant Function is proposed in this study which makes the traditional function more sensitive to the important instances. This way, the profit that can be obtained from a fraud/legitimate classifier is maximized. Experimental results confirm that Modified Fisher Discriminant could eventuate more profit.
The aim of bankruptcy prediction in the areas of data mining and machine learning is to develop an effective model which can provide the higher prediction accuracy. In the prior literature, various classification techniques have been developed and studied, in/with which classifier ensembles by combining multiple classifiers approach have shown their outperformance over many single classifiers. However, in terms of constructing classifier ensembles, there are three critical issues which can affect their performance. The first one is the classification technique actually used/adopted, and the other two are the combination method to combine multiple classifiers and the number of classifiers to be combined, respectively. Since there are limited, relevant studies examining these aforementioned disuses, this paper conducts a comprehensive study of comparing classifier ensembles by three widely used classification techniques including multilayer perceptron (MLP) neural networks, support vector machines (SVM), and decision trees (DT) based on two well-known combination methods including bagging and boosting and different numbers of combined classifiers. Our experimental results by three public datasets show that DT ensembles composed of 80-100 classifiers using the boosting method perform best. The Wilcoxon signed ranked test also demonstrates that DT ensembles by boosting perform significantly different from the other classifier ensembles. Moreover, a further study over a real-world case by a Taiwan bankruptcy dataset was conducted, which also demonstrates the superiority of DT ensembles by boosting over the others.
The restaurant industry has been facing tough challenges because of the recent economic turmoil. Although different industries face different levels of competition and therefore the likelihood of financial distress can differ for firms in different industries, scant attention has been paid to predicting restaurant financial distress. The primary objective of this paper is to examine the key financial distress factors for publicly traded U.S. restaurants for the period from 1988 to 2010 using decision trees (DT) and AdaBoosted decision trees. The AdaBoosted DT model for the entire dataset revealed that financially distressed restaurants relied more heavily on debt; and showed lower rates of increase of assets, lower net profit margins, and lower current ratios than non-distressed restaurants. A larger proportion of debt in the capital structure ruined restaurants' financial structure and the inability to pay their drastically increased debt exposed restaurants to financial distress. Additionally, a lack of capital efficiency increased the possibility of financial distress. We recommend the use of the AdaBoosted DT model as an early warning system for restaurant distress prediction because the AdaBoosted DT model demonstrated the best prediction performance with the smallest error in overall and type I error rates. The results of two subset models for full-service and limited-service restaurants indicated that the segments had slightly different financial risk factors.
We develop a model of neural networks to study the bankruptcy of U.S. banks, taking into account the specific features of the recent financial crisis. We combine multilayer perceptrons and self-organizing maps to provide a tool that displays the probability of distress up to three years before bankruptcy occurs. Based on data from the Federal Deposit Insurance Corporation between 2002 and 2012, our results show that failed banks are more concentrated in real estate loans and have more provisions. Their situation is partially due to risky expansion, which results in less equity and interest income. After drawing the profile of distressed banks, we develop a model to detect failures and a tool to assess bank risk in the short, medium and long term using bankruptcies that occurred from May 2012 to December 2013 in U.S. banks. The model can detect 96.15% of the failures in this period and outperforms traditional models of bankruptcy prediction.
In classification or prediction tasks, data imbalance problem is frequently observed when most of instances belong to one majority class. Data imbalance problem has received considerable attention in machine learning community because it is one of the main causes that degrade the performance of classifiers or predictors. In this paper, we propose geometric mean based boosting algorithm (GMBoost) to resolve data imbalance problem. GMBoost enables learning with consideration of both majority and minority classes because it uses the geometric mean of both classes in error rate and accuracy calculation. To evaluate the performance of GMBoost, we have applied GMBoost to bankruptcy prediction task. The results and their comparative analysis with AdaBoost and cost-sensitive boosting indicate that GMBoost has the advantages of high prediction power and robust learning capability in imbalanced data as well as balanced data distribution.
A lot of bankruptcy forecasting model has been studied. Most of them uses corporate finance data and is intended for general companies. It may not appropriate for forecasting bankruptcy of construction companies which has big liquidity. It has a different capital structure, and the model to judge the financial risk of general companies can be difficult to apply the construction companies. The existing studies such as traditional Z-score and bankruptcy prediction using machine learning focus on the companies of nonspecific industries. The characteristics of companies are not considered at all. In this paper, we showed that AdaBoost (adaptive boosting) is an appropriate model to judge the financial risk of Korean construction companies. We classified construction companies into three groups - large, middle, and small based on the capital of a company. We analyzed the predictive ability of the AdaBoost and other algorithms for each group of companies. The experimental results showed that the AdaBoost has more predictive power than others, especially for the large group of companies that has the capital more than 50 billion won.
Corporate going-concern opinions are not only useful in predicting bankruptcy but also provide some explanatory power in predicting bankruptcy resolution. The prediction of a firm's ability to remain a going concern is an important and challenging issue that has served as the impetus for many academic studies over the last few decades. Although intellectual capital (IC) is generally acknowledged as the key factor contributing to a corporation's ability to remain a going concern, it has not been considered in early prediction models. The objective of this study is to increase the accuracy of going-concern prediction by using a hybrid random forest (RF) and rough set theory (RST) approach, while adopting IC as a predictive variable. The results show that this proposed hybrid approach has the best classification rate and the lowest occurrence of Types I and II errors, and that IC is indeed valuable for going-concern prediction.
We investigated the performance of parametric and non-parametric methods concerning the in-sample pricing and out-of-sample prediction performances of index options. Comparisons were performed on the KOSPI 200 Index options from January 2001 to December 2010. To verify the statistical differences between the compared methods, we tested the following null hypothesis: two series of forecasting errors have the same mean-squared value. The experimental study reveals that non-parametric methods significantly outperform parametric methods on both in-sample pricing and out-of-sample pricing. The outperforming non-parametric method is statistically different from the other models, and significantly different from the parametric models. The Gaussian process model delivers the most outstanding performance in forecasting, and also provides the predictive distribution of option prices.
With the recent financial crisis and European debt crisis, corporate bankruptcy prediction has become an increasingly important issue for financial institutions. Many statistical and intelligent methods have been proposed, however, there is no overall best method has been used in predicting corporate bankruptcy. Recent studies suggest ensemble learning methods may have potential applicability in corporate bankruptcy prediction. In this paper, a new and improved Boosting, FS-Boosting, is proposed to predict corporate bankruptcy. Through injecting feature selection strategy into Boosting, FS-Booting can get better performance as base learners in FS-Boosting could get more accuracy and diversity. For the testing and illustration purposes, two real world bankruptcy datasets were selected to demonstrate the effectiveness and feasibility of FS-Boosting. Experimental results reveal that FS-Boosting could be used as an alternative method for the corporate bankruptcy...
Seasonality effects and empirical regularities in financial data have been well documented in the financial economics literature for over seven decades. This paper proposes an expert system that uses novel machine learning techniques to predict the price return over these seasonal events, and then uses these predictions to develop a profitable trading strategy. While simple approaches to trading these regularities can prove profitable, such trading leads to potential large drawdowns (peak-to-trough decline of an investment measured as a percentage between the peak and the trough) in profit. In this paper, we introduce an automated trading system based on performance weighted ensembles of random forests that improves the profitability and stability of trading seasonality events. An analysis of various regression techniques is performed as well as an exploration of the merits of various techniques for expert weighting. The performance of the models is analysed using a large sample of stocks from the DAX. The results show that recency-weighted ensembles of random forests produce superior results in terms of both profitability and prediction accuracy compared with other ensemble techniques. It is also found that using seasonality effects produces superior results than not having them modelled explicitly.
This paper extends the macroeconomic frailty model to include sectoral frailty factors that capture default correlations among firms in a similar business. We estimate sectoral and macroeconomic frailty factors and their effects on default intensity using the data for Japanese firms from 1992 to 2010. We find strong evidence for the presence of sectoral frailty factors even after accounting for the effects of observable covariates and macroeconomic frailty on default intensity. The model with sectoral frailties performs better than that without. Results show that accounting for the sources of unobserved sectoral default risk covariations improves the accuracy of default probability estimation.
Consumer credit scoring is often considered a classification task where clients receive either a good or a bad credit status. Default probabilities provide more detailed information about the creditworthiness of consumers, and they are usually estimated by logistic regression. Here, we present a general framework for estimating individual consumer credit risks by use of machine learning methods. Since a probability is an expected value, all nonparametric regression approaches which are consistent for the mean are consistent for the probability estimation problem. Among others, random forests (RF), k-nearest neighbors (kNN), and bagged k-nearest neighbors (bNN) belong to this class of consistent nonparametric regression approaches. We apply the machine learning methods and an optimized logistic regression to a large dataset of complete payment histories of short-termed installment credits. We demonstrate probability estimation in Random Jungle, an RF package written in C++ with a generalized framework for fast tree growing, probability estimation, and classification. We also describe an algorithm for tuning the terminal node size for probability estimation. We demonstrate that regression RF outperforms the optimized logistic regression model, kNN, and bNN on the test data of the short-term installment credits.
Decision tree (DT) is one of the most popular classification algorithms in data mining and machine learning. However, the performance of DT based credit scoring model is often relatively poorer than other techniques. This is mainly due to two reasons: DT is easily affected by (1) the noise data and (2) the redundant attributes of data under the circumstance of credit scoring. In this study, we propose two dual strategy ensemble trees: RS-Bagging DT and Bagging-RS DT, which are based on two ensemble strategies: bagging and random subspace, to reduce the influences of the noise data and the redundant attributes of data and to get the relatively higher classification accuracy. Two real world credit datasets are selected to demonstrate the effectiveness and feasibility of proposed methods. Experimental results reveal that single DT gets the lowest average accuracy among five single classifiers, i.e., Logistic Regression Analysis (LRA), Linear Discriminant Analysis (LDA), Multi-layer Perceptron (MLP) and Radial Basis Function Network (RBFN). Moreover, RS-Bagging DT and Bagging-RS DT get the better results than five single classifiers and four popular ensemble classifiers, i.e., Bagging DT, Random Subspace DT, Random Forest and Rotation Forest. The results show that RS-Bagging DT and Bagging-RS DT can be used as alternative techniques for credit scoring.
Bagging predictors is a method for generating multiple versions of a predictor and using these to get an aggregated predictor. The aggregation averages over the versions when predicting a numerical outcome and does a plurality vote when predicting a class. The multiple versions are formed by making bootstrap replicates of the learning set and using these as new learning sets. Tests on real and simulated data sets using classification and regression trees and subset selection in linear regression show that bagging can give substantial gains in accuracy. The vital element is the instability of the prediction method. If perturbing the learning set can cause significant changes in the predictor constructed, then bagging can improve accuracy.