Figure 2 - uploaded by Mahdi Salehi
Content may be subject to copyright.
Source publication
The current study seeks to provide a new solution for evaluation of banking system customers risk by integrating different scientific methodology. Evaluation of banking system customers risk in Iranian banks relies on experts judgment and fingertip rule. This type of evaluation resulted in high rate of postponed claims; therefore, designing new int...
Contexts in source publication
Context 1
... fitting means that the model in the learning process, amount of training error and generalization of model for new data decreases. Figure 2 demonstrates a sample model of over fitting (Liao, 1999;Back and Back, 1995). ...
Context 2
... step by step methods similar to regression model are used to determine neuron's number. In any case, using network with greatly/numerous neuron is efficient than rigid network with constant neurons (Rumelhart et al., 2006). In assay, we used step by step method to detect the number of neurons. ...
Similar publications
Customer choice and awareness have been increasing tremendously during this decade because of more transparency in the economy, the advent of information technology, media revolution and besides hectic competition for resources among banks. Service quality is a judgmental issue relating to the different between an individual’s expectation of a serv...
Citations
... This means that the networks are capable of adapting to arbitrary and unknown functional forms, with an arbitrarily specified degree of precision. Neural network and logistic regression in forecasting customer credit risk have same efficiency [5]. The important roles of ANN in financial application are pattern recognition, classification and time series forecasting [6]. ...
... They concluded in the light of experimental results that the ANN system outperforms the SVM in terms of accuracy (85.305% for ANN and 84% for SVM) and the minimum learning time. Salehi & Mansoury (2011) explored the effectiveness of ANN and LR in the prediction of credit risk. They claimed that both models are similarly efficient. ...
This work tries to determine the probability of default as a tool to measure credit risk in a Tunisian commercial bank. A scoring model was built according to the traditional technique of logistic regression (LR), and artificial intelligence techniques i.e. artificial neural networks (ANN) and support vector machines (SVM). Then a comparison was made between these models using performance metrics such as the confusion matrix and the area under the ROC curve (AUC) in order to identify the most efficient model. Our results show that the Radial Basis Function kernel SVM was the most performing method in terms of accuracy, sensitivity and specificity with the least error rates. Thus, in the Tunisia context, this model is worth implementing in banking institutions in order to improve their credit risk management measures to monitor and control credit.
... However, the mono-method-based models cannot take into consideration all demands of model robustness, prediction accuracy and variables' interpretation. Therefore, combination optimization of multiple methods has become an important developing for credit assessment [11]. Logistic regression-support vector machine combination (LR-SVM) model could extract significant variables, simplify the model structure, reduce the computational complexity and improve the model efficiency [12]. ...
... Tsai and Wu (2008) demonstrated that neural networks with a back-propagation algorithm are the most popular tools in making financial decisions, due to their unique capability in data classification. Salehi and Mansouri (2011) investigated the performance of logistic regression and neural networks in predicting customers' credit scores, which confirmed that there exists an equilibrium between these methods. Elter and Yaseen (2010) addressed the importance of financial neural networks in pattern identification as well as time series classification and prediction. ...
The credit scoring system is one of the most significant credit risk control tools in the banking industry. Usually, the existing credit scoring models classify customers into "good credit" and "bad credit" groups. In this study, a novel model is proposed to classify corporate client accounts into four groups - good credit, past due, overdue and doubtful - according to the definitions of the Central Bank of the Islamic Republic of Iran. This model enables lenders to develop specific policies for credit granting by predicting the solvency and insolvency rates of their corporate clients. For validation, the proposed model, trained by the hybrid approach of a self-organizing map and radial basis function (RBF) neural network, is compared with a single-step, four-class classification model and a model trained by support vector machines. The results show that the proposed model trained by the hybrid approach of a self-organizing map and RBF neural network outperforms the existing methods in terms of its final accuracy with regard to the four classes at the test stage.
... But such results have not been confirmed by additional research (Desai et al., 1996; Bell, 1997; Laitinen and Kankaanpaa, 1999; Salehi, 2011). Desai et al. (1996) compare the performance of Multilayer perceptron, LR and linear discriminant analysis. ...
... Bell (1997) also used LR and ANN to predict commercial bank failure and concludes that both models are equally efficient and neither can be considered as a dominant predictive model. Salehi (2011) compared the accuracy of these techniques to identify bad and good debtors after the loan was obtained, using 11 input variables relating to the company's financial position and the industry it belonged to. ANN and LR have similar prediction efficiency. ...
In this paper, we establish a general framework for explaining the functioning of Artificial Neural Networks (ANNs) in binomial classification. We compare this approach with one of the conventional techniques, namely logistic regression (LR) to predict credit rationing (CR) decisions. We use leave-one-out cross validation to ensure the robustness of the two classifiers. Based on data from a survey of 246 French Small and Medium Enterprises (SMEs), we highlight the imbalanced learning issue as the fact that the class distributions of rationed and non-rationed firms are skewed. We apply a comprehensive approach using a combination of oversampling and under sampling methods called ROS/RUS. We balance the data in the learning phase and compare the results with those obtained on the original data. We estimate the classifiers’ performance based on the confusion matrix, the accuracy rate with paired t-tests, the ROC curve and the Area under the ROC curve (AROC). Our study can contribute to the choice of the best approach in CR prediction for small businesses.
... Multilayer feed forward networks are a class of universal approximation (Hornik, Stinchcombe & White, 1989). The ANN models have a high predictive power (Steiner, Neto, Soma, Shimizu & Nievola, 2006 ). Networks are nonparameters model which is an arbitrarily specified degree of precision. ...
The aim of this paper is to examine the efficiency of two credit risk modeling (CRM) to predict the credit risk of commercial Iranian banks: (1) Logistic regression model (LRM); (2) Artificial neural networks (ANNs). The calculations have been done by using SPSS and MATLAB software. Number of samples was 316 and 5 dependent variables. The results showed that, artificial neural network is more proper to identify bad customers in commercial bank. The major contribution of this paper is specifying the most important determinants for rating of customers in Iran's banking sector.
... Use of modern indicators in addition to traditional financial ratio indicators provides considerable improvement in forecast accuracy (Atiya, 2001). Salehi and Mansoury (2011) investigate the efficiency of neural network and logistic regression in forecasting customer credit risk. They state that both models have same efficiency. ...
In many studies, the relationship between development of financial markets and economic growth has been proved. Credit risk is one of problems which banks are faced with it while doing their tasks. Credit risk means the probability of non-repayment of bank financial facilities granted to investors. If the credit risk decreases, banks will be more successful in performing their duties and have greater effect on economic growth of the country. Credit rating of customers and identifying good and bad customers, help banks lend to their good payers and hereby, they reduce probability of non-repayment. This paper aims to identify classification criteria for good customers and bad customers in Iranian banks. This study can classified in applied studies group and the research strategy is descriptive. Artificial neural network technique is used for financial facilities applicants' credit risk measurement and the calculations have been done by using SPSS and MATLAB software. Number of samples was 497 and 18 variables were used to identify good customers from bad customers. The results showed that, individual loan frequency and amount of loan had most important effect and also status of customer's bank account, history of customer relationship with bank and received services had least important effect in identifying classification criteria of good and bad customers. The major contribution of this paper is specifying the most important determinants for rating of customers in Iran's banking sector.
The purpose of this study is to identify important variables that influence on credit risk. Statistical analysis was used. In order to achieve the purpose of this research, a frame of references has been constructed based on a wide literature review. The calculations have been done by using SPSS 18 software. Number of samples was 90 and 5 dependent variables. The achieved results indicate the relation between credit risk and independent variables were considered. The major contribution of this paper is specifying the most important determinants for rating of customers in Iran's banking sector.
