Predictive Modeling - Science topic

As engineering and material science increasingly adopt data-driven approaches, I am intrigued by the rapid advancements in supervised Machine Learning (ML) and Deep Learning (DL). These tools have proven transformative, but I am eager to explore a question at the heart of innovation in this space:

Specifically, I am inviting AI researchers and practitioners to share insights into:

In my concrete durability and sustainability research, I aim to leverage AI tools for academic insights and to provide actionable solutions for designing better, longer-lasting materials. To achieve this, it is critical to understand and integrate the most cutting-edge tools available today. For example:

I am curious to hear from the community:

This is not just a call to share tools or techniques. It is an invitation to discuss how these advancements can be meaningfully applied to solve practical problems in engineering and beyond. I look forward to hearing about your experiences, discoveries, and perspectives on the evolving role of AI in research and practice.

Let’s connect and explore how we can drive innovation with purpose and impact!

Artificial Intelligence (AI) is rapidly transforming industries across the globe, from healthcare and finance to education and entertainment. In particular, the fields of research and writing have been significantly impacted by AI’s capabilities. AI tools are helping researchers and writers optimize their workflows, boost productivity, and improve the quality of their work. Whether it's generating ideas, improving grammar, or assisting in data analysis, AI has revolutionized how research papers are written, edited, and managed.

In academic writing, AI tools are particularly beneficial for both novice and experienced researchers. These tools can assist in various stages of writing, from brainstorming topics to generating drafts, all the way to final editing. By automating repetitive tasks, AI allows researchers to focus more on the critical aspects of their work, such as data analysis, hypothesis formulation, and scholarly discourse. In this article, we will explore the role of AI in writing research papers and the tools that are transforming the research process.

One of the most popular and impactful uses of AI in academic writing is through AI-powered writing tools. These tools help researchers improve their writing by offering grammar and style suggestions, sentence structure enhancements, and readability improvements. Some of the most widely used AI writing tools include Grammarly, Hemingway, and ProWritingAid.

In addition to improving grammar and style, these AI tools can also assist researchers in generating ideas and outlining papers. For instance, AI-powered systems like QuillBot can help researchers paraphrase text, generate content ideas, and even suggest keywords or phrases related to specific research topics. Some advanced AI systems can also help in suggesting relevant literature by analysing existing research and identifying gaps in the literature that can be explored in new papers.

A literature review is an essential part of any research paper, dissertation, or thesis. It involves reviewing existing research to identify gaps, trends, and key findings that inform the research project. AI tools have significantly simplified the process of conducting literature reviews by enabling researchers to analyse vast databases of academic papers and summarize relevant findings.

By automating the literature review process, AI tools can save researchers hours of manual work, allowing them to focus on synthesizing the findings and developing new research questions.

AI is not only transforming the writing and review process but also plays a significant role in managing the research process itself. Tools like Zotero, EndNote, and Mendeley are widely used by researchers to manage citations, references, and research materials efficiently.

AI-powered tools help researchers save time by automating tedious tasks like managing references, analysing data, and formatting papers. This allows them to focus on higher-level tasks like formulating hypotheses, designing experiments, and interpreting results.

AI tools are also making strides in assisting researchers with the writing process itself. OpenAI’s GPT-3, for example, is an AI model capable of generating human-like text. GPT-3 can assist researchers by generating drafts, suggesting sentence structures, or even creating content for specific sections of a research paper.

However, researchers should be cautious when using AI tools like GPT-3. While these tools can be incredibly helpful in drafting content and improving readability, they may not fully understand the nuances of academic writing, particularly in specialized fields. Researchers should always review the generated content and ensure that it aligns with the specific requirements of their research.

The integration of AI in the research and writing process has revolutionized how academic work is completed. AI tools offer significant benefits, including improved writing quality, faster research, better organization, and enhanced productivity. Tools like Grammarly, Hemingway, Zotero, and GPT-3 can help researchers with everything from generating content to managing references, performing data analysis, and writing literature reviews.

While AI tools can greatly enhance the efficiency and quality of academic writing, it is important for researchers to retain control over their work. AI should be seen as an assistant, not a replacement, for human intelligence. The integrity of research and writing still depends on the researcher’s expertise, judgment, and understanding of the subject matter.

Researchers are encouraged to embrace AI tools for the efficiency they offer but to always ensure that their work maintains its originality, rigour, and scholarly value. If you're looking for personalized support in your research and writing journey, consider reaching out to Hamza Omullah for expert guidance, coaching, and consulting services.

Explore AI writing tools today and share your experiences with them! For personalized research and writing assistance, feel free to reach out to Hamza at HAMNIC Solutions.

#ResearchConsultant #WritingSupport #AcademicWriting #ResearchHelp #ContentCreation #ProfessionalWriting #ThesisWriting #DissertationHelp #AcademicSuccess #ClientSuccess #hamnicwriting #hamnicsolutions #AI

The Schrödinger equation, as foundational as it is in quantum mechanics, fails to adequately describe the true nature of quantum particle motion, as demonstrated by my recent research ( DOI: 10.9790/4861-1505012633). This raises the critical question: What alternative frameworks can we use to better understand quantum mechanics? Given that the current models have been proven insufficient, it becomes crucial to explore different ways to model the behavior of particles. Can we ever truly predict the motion of quantum particles accurately, considering our limited understanding of both the inner workings of matter (with medical science only scratching the surface of human biology) and the physical universe (with only 7% of the observable matter understood)? The complexities of quantum motion are so vast, and our scientific knowledge so constrained, that predicting exact particle states may well remain unattainable.

This query invites the ResearchGate community to propose alternatives to the Schrödinger equation. What would a new model look like, and can we develop a formula that predicts quantum motion in a way that better aligns with the complexities and limitations of our current knowledge? Given the undeniable limitations of modern science, can we ever predict the precise mechanics of a particle? Your insights and suggestions are welcome.

“Does anyone have any reference or methodology for projecting an Urban Heat Island (UHI) layer into the future using GIS? I am interested in techniques that combine satellite data and predictive modeling tools to project changes in urban surface temperature. I welcome any input or related studies.”

Att:

Jhonsy O. Silva-López

What is the basis for determining both H and K in the ARIMA models in determining the best predictive model?

I am looking for collaborators to work on a research project based on prediction models of tomato yield losses caused by tomato brown rugose fruit virus (ToBRFV). The study is being carried out in Mexico.

More information by this way, or via mail.

I am developing a predictive model for a water supply network that involves 20 influencing points. However, I only have historical data for 10 out of these 20 points.

I would like to know how to approach building a predictive model given this incomplete dataset. Specifically:

AI-driven technologies offer previously unheard-of capabilities to process enormous volumes of data, extract insightful knowledge and improve predictive models, according to the UN’s World Meteorological Organization (WMO). That means improved modelling and predicting climate change patterns that can help communities and authorities to draft effective adaptation and mitigation strategies.

source: https://news.un.org/en/story/2023/11/1143187

As a natural ressources manager, I am exploring the viability of using slope as the primary indicator for assessing soil susceptibility to erosion. While slope is a significant factor, I am keen to understand its limitations and the additional variables that must be considered for a comprehensive evaluation.

I am looking for detailed insights, experiences, and references that can deepen the understanding of this complex issue and contribute to more robust soil erosion assessment methodologies.

Project is to conduct a comprehensive analysis of online retail sales trends and customer behaviour using data collected from online sources. The project aims to derive actionable insights for optimising marketing strategies, improving customer experience, and enhancing overall business performance for e-commerce businesses. It will include the analysis of online retail data sourced from various e-commerce platforms, APIs, or web scraping techniques. The focus will be on understanding the factors influencing sales, identifying patterns in customer behaviour, and predicting future sales trends. Include customer segmentation and predictive modelling

I need any software that help me to calculate Harrell’s C-index for concordance of predictive model

Please

I am using the CERES rice model in DSSAT to model CA rice production. The model is unable to predict lodging and damage from pests and diseases. I am modeling a cultivar that has been fairly extensively studied in California. The wealth of field trial data available for this cultivar allows a fairly accurate sense of losses due to lodging and disease. It seems logical that I could apply these loss rates to the total yield predicted by the model to get a more accurate yield output. However, I'm struggling to determine whether there is precedence for this kind of post-model adjustment? 

I understand that there are many machine learning predictive model, I just want to know the best model that can be used to predict customer behavior with the least amount of error .

ROC curve analysis to detect the best cut-off of the independent variables in the prediction models

We are working on a predictive model using Lasso to evaluate the validation cohort using publicly available data.

1. ROC-AUC 0.800 (95%CI 0.551-1.000)

2. ROC-AUC 0.750 (95%CI 0.350-1.000)

3. ROC-AUC 0.575 (95%CI 0.317-1.000)

I am wondering how to describe the evaluation of (2): please let me know if I can describe (1) and (2) as being well predicted statistically. Since the lower 95% CI limit is 0.35, despite the high ROC-AUC of 0.75, shouldn't we describe the evaluation of the model positively? thank you so much for your support.

how to do calibration curves for the prediction models in SPSS, i think it is important in addition to the external validation

Dear ResearchGate Community,

I am in the process of developing a machine learning (ML) predictive model to study the degradation and performance of corroded prestressed steel in concrete structures. The objective is to utilize advanced ML algorithms to predict the long-term effects of corrosion on the mechanical properties of prestressed steel.

For this purpose, I am seeking access to the CEB-FIP database or any similar repository containing comprehensive data on corroded prestressed steel. This data is crucial for training and validating the ML models to ensure accurate predictions. I am particularly interested in datasets that include corrosion rates, mechanical property degradation, fatigue life, and other parameters critical to the structural performance of these materials.

If anyone has access to the CEB-FIP database or knows of similar databases that could serve this research purpose, I would greatly appreciate your assistance in gaining access.

Your support would be invaluable in furthering our understanding of material behavior in civil engineering and developing robust tools for predicting structural integrity.

I am open to collaborations and would be keen to discuss potential joint research initiatives that explore the application of machine learning in civil and structural engineering.

Thank you for your time and consideration. I look forward to any possible assistance or collaboration from the community.

Best regards,

M. Kovacevic

I would like to know if there is a program or website (platform) that aims to create predictive models just by providing the database? That is, is there a free program with the standard molecular modeling (statistical tools) steps implemented and is easy to use?

From my understanding, the baseline hazard or baseline survival function is unkown because cox regression is semi-parametric model. So why and how can we use it as a prediction model, for example using it to predict the 10 years survival probability.

Hi i have estimated an armax model using python sippy library. The estimation gives me two transfer functions H and G. How can I combine them into a single one to predict model output for new input u(t)or to compute unit step response? I thought to somehow derive state space representation maybe...

Hi

I have a non-linear equation

I want to convert it into a linear equation using the predictive model method

Can anyone help?

and then optimize with linear programming method

please send an e-mail

I want to develop a predictive model using random forest, SVM, and decision tree algorithms. How can I estimate my sample size? Is there any formula?

Is it possible to build a highly effective, multi-faceted, intelligent forecasting system for future financial and economic crises based on artificial intelligence technology in combination with Data Science analytics, Big Data Analytics, Business Intelligence and/or other Industry 4.0 technologies as part of a forecasting system for complex, multi-faceted economic processes in such a way as to reduce the scale of the impact of the paradox of a self-fulfilling prediction and to increase the scale of the paradox of not allowing a predicted crisis to occur due to pre-emptive anti-crisis measures applied?

What do you think about the involvement of artificial intelligence in combination with Data Science, Big Data Analytics, Business Intelligence and/or other Industry 4.0 technologies for the development of sophisticated, complex predictive models for estimating current and forward-looking levels of systemic financial, economic risks, debt of the state's public finance system, systemic credit risks of commercially operating financial institutions and economic entities, forecasting trends in economic developments and predicting future financial and economic crises?

Research and development work is already underway to teach artificial intelligence to 'think', i.e. the conscious thought process realised in the human brain. The aforementioned thinking process, awareness of one's own existence, the ability to think abstractly and critically, and to separate knowledge acquired in the learning process from its processing in the abstract thinking process in the conscious thinking process are just some of the abilities attributed exclusively to humans. However, as part of technological progress and improvements in artificial intelligence technology, attempts are being made to create "thinking" computers or androids, and in the future there may be attempts to create an artificial consciousness that is a digital creation, but which functions in a similar way to human consciousness. At the same time, as part of improving artificial intelligence technology, creating its next generation, teaching artificial intelligence to perform work requiring creativity, systems are being developed to process the ever-increasing amount of data and information stored on Big Data Analytics platform servers and taken, for example, from selected websites. In this way, it may be possible in the future to create "thinking" computers, which, based on online access to the Internet and data downloaded according to the needs of the tasks performed and processing downloaded data and information in real time, will be able to develop predictive models and specific forecasts of future processes and phenomena based on developed models composed of algorithms resulting from previously applied machine learning processes. When such technological solutions become possible, the following question arises, i.e. the question of taking into account in the built intelligent, multifaceted forecasting models known for years paradoxes concerning forecasted phenomena, which are to appear only in the future and there is no 100% certainty that they will appear. Well, among the various paradoxes of this kind, two particular ones can be pointed out. One is the paradox of a self-fulfilling prophecy and the other is the paradox of not allowing a predicted crisis to occur due to pre-emptive anti-crisis measures applied. If these two paradoxes were taken into account within the framework of the intelligent, multi-faceted forecasting models being built, their effect could be correlated asymmetrically and inversely proportional. In view of the above, in the future, once artificial intelligence has been appropriately improved by teaching it to "think" and to process huge amounts of data and information in real time in a multi-criteria, creative manner, it may be possible to build a highly effective, multi-faceted, intelligent forecasting system for future financial and economic crises based on artificial intelligence technology, a system for forecasting complex, multi-faceted economic processes in such a way as to reduce the scale of the impact of the paradox of a self-fulfilling prophecy and increase the scale of the paradox of not allowing a predicted crisis to occur due to pre-emptive anti-crisis measures applied. In terms of multi-criteria processing of large data sets conducted with the involvement of artificial intelligence, Data Science, Big Data Analytics, Business Intelligence and/or other Industry 4. 0 technologies, which make it possible to effectively and increasingly automatically operate on large sets of data and information, thus increasing the possibility of developing advanced, complex forecasting models for estimating current and future levels of systemic financial and economic risks, indebtedness of the state's public finance system, systemic credit risks of commercially operating financial institutions and economic entities, forecasting economic trends and predicting future financial and economic crises.

In view of the above, I address the following questions to the esteemed community of scientists and researchers:

What do you think about this topic?

What is your opinion on this subject?

Please respond,

I invite you all to discuss,

Thank you very much,

Warm regards,

Dariusz Prokopowicz

Ongoing technological progress is increasing the technical possibilities of both conducting research, collecting and assembling large amounts of research data and their multi-criteria processing using ICT information technologies and Industry 4.0. Before the development of ICT information technologies, IT tools, personal computers, etc. in the second half of the 20th century as part of the 3rd technological revolution, computerised, semi-automated processing of large data sets was very difficult or impossible. As a result, the building of multi-criteria, multi-article, big data and information models of complex macro-process structures, simulation models, forecasting models was limited or practically impossible. However, the technological advances made in the current fourth technological revolution and the development of Industry 4.0 technology have changed a lot in this regard. The current fourth technological revolution is, among other things, a revolution in the improvement of multi-criteria, computerised analytical techniques based on large data sets. Industry 4.0 technologies, including Big Data Analytics technology, are used in multi-criteria processing, analysing large data sets. Artificial Intelligence (AI) can be useful in terms of scaling up the automation of research processes and multi-faceted processing of big data obtained from research.

The technological advances taking place are contributing to the improvement of computerised analytical techniques conducted on increasingly large data sets. The application of the technologies of the fourth technological revolution, including ICT information technologies and Industry 4.0 in the process of conducting multi-criteria analyses and simulation and forecasting models conducted on large sets of information and data increases the efficiency of research and analytical processes. Increasingly, in research conducted within different scientific disciplines and different fields of knowledge, analytical processes are carried out, among others, using computerised analytical tools including Big Data Analytics in conjunction with other Industry 4.0 technologies.

When these analytical tools are augmented with Internet of Things technology, cloud computing and satellite-implemented sensing and monitoring techniques, opportunities arise for real-time, multi-criteria analytics of large areas, e.g. nature, climate and others, conducted using satellite technology. When machine learning technology, deep learning, artificial intelligence, multi-criteria simulation models, digital twins are added to these analytical and research techniques, opportunities arise for creating predictive simulations for multi-factor, complex macro processes realised in real time. Complex, multi-faceted macro processes, the study of which is facilitated by the application of new ICT information technologies and Industry 4.0, include, on the one hand, multi-factorial natural, climatic, ecological, etc. processes and those concerning changes in the state of the environment, environmental pollution, changes in the state of ecosystems, biodiversity, changes in the state of soils in agricultural fields, changes in the state of moisture in forested areas, environmental monitoring, deforestation of areas, etc. caused by civilisation factors. On the other hand, complex, multifaceted macroprocesses whose research processes are improved by the application of new technologies include economic, social, financial, etc. processes in the context of the functioning of entire economies, economic regions, continents or in global terms.

Year on year, due to technological advances in ICT, including the use of new generations of microprocessors characterised by ever-increasing computing power, the possibilities for increasingly efficient, multi-criteria processing of large collections of data and information are growing. Artificial intelligence can be particularly useful for the selective and precise retrieval of specific, defined types of information and data extracted from many selected types of websites and the real-time transfer and processing of this data in database systems organised in cloud computing on Big Data Analytics platforms, which would be accessed by a system managing a built and updated model of a specific macro-process using digital twin technology. In addition, the use of supercomputers, including quantum computers characterised by particularly large computational capacities for processing very large data sets, can significantly increase the scale of data and information processed within the framework of multi-criteria analyses of natural, climatic, geological, social, economic, etc. macroprocesses taking place and the creation of simulation models concerning them.

In view of the above, I address the following question to the esteemed community of scientists and researchers:

What do you think about this topic?

What is your opinion on this subject?

Please respond,

I invite you all to discuss,

Thank you very much,

Warm regards,

The above text is entirely my own work written by me on the basis of my research.

I have not used other sources or automatic text generation systems such as ChatGPT in writing this text.

Dariusz Prokopowicz

I would like to construct principal components from 30+ financial ratios for a predictive model. I would then use logistic regression and support vector machines for the predictions, conditioned on the principal components. I have panel data with T << N. For PCA, the data should be iid. I am concerned that the time series are not independent. I have only 10 years of data, which precludes most time series statistical testing. I have seen several peer-reviewed papers constructing principal components using panel data, but the potential problems with using panel data for PCA are not discussed in the papers. The papers all seem to use the standard PCA approach one would normally use with a cross-section but with panel data. I have researched several means of doing a PCA with time series and a couple that use panel data as one of several examples of a general (usually very complicated) PCA procedure, but nothing seems to "fit the bill" for my standard panel dataset. I would greatly appreciate some direction as to where I might go to look for an adequate procedure or suggestions for a procedure that could possibly work. I am an applied econometrician, and it would be difficult for me to translate a complex procedure into code. So, ideally, I would like to find a procedure for which there is existing code (I use SAS and R). Thanks in advance for any insights provided,

AI aids in climate change mitigation through various applications, such as optimizing renewable energy systems, analyzing climate data for predictive modeling, and enabling more efficient resource management, leading to reduced greenhouse gas emissions and more sustainable practices.

Hi. I'm planning to conduct a multinomial logistic regression analysis for my predictive model (3 outcome categories). How can I estimate the sample size? I believe the EPV rule of thumb is suitable only for binary outcomes. Is there any formula/software that I can use?

I already tried using pvisgam (itsadug library) and although it does include the color key (zlim code) this only works for partial effects. Thanks in advance.

How does R programming support machine learning and predictive modelling in research applications? Can you highlight some popular R packages and algorithms that researchers can use for building and evaluating predictive models?

kindly want to know all model for financial model other yhen altman 1968- beaven 1966

kinda 1980- sherrod 1987 - --

how else

Illustrate the comparative analysis of VAN Method, Poison Regression or any other methods available?

Can you explain it or recommend some papers?

(1) For classification problem, we can use recall rate, precision rate and F1 score, AUC, balanced accuracy(although I don't know what is mean/ max balanced accuracy) in same time to compare different classifiers.

but I haven't see any paper to evaluate different numerical prediction models (forgive my ignorance).

(2) I only know r-squared, MAE, MAPE and RMSE could be used to assess models, but how to combine them to evaluate in a certain case, should add any other metrics?

(3) Could you help explain max/mean balanced accuracy, mean/max recall, mean/max accuracy, mean/max AUC, mean/max F1 if you are free. Why use mean or max, aren't these values ​​specific values? Why are they divided into mean and max?

There are lots of metrics such r-square, MSE, MAE, MAPE and others. Some paper use r-square combining RMSE, some use EVS+MAPE+t-test+NSE, so if I want to evaluate a price prediction model or a conversion rate prediction model, what combination should I choose?

I am working on a project to develop a predictive model to predict the probability of tax revenue (continous value). Any previous model or experience in the field?

I am working on a project to develop a predictive model to predict the probability of student success in completing a program. Any previous model or experience in the field?

I would like to know what regression algorithm works best with complex data with good accuracy and RMSE score.

Probabilistic modelling, Forecasting, Renewable energy uncertainty prediction, ARIMA

Considering that deep learning models can automatically learn features from data, do they need other special feature engineering techniques to attain high accuracy given that it is challenging to extract relevant features using most feature engineering tools and that deep neural networks need more processing power and time to learn well when dealing with complex data?

If needed, in what context of application will doing this be required and how would this impact the model performance?

Contrary to the above, for better model performance what would be your recommendation of the most suited type of deep learning algorithm to be implemented for Image Recognition, Natural Language Processing (NLP), and different types of Predictive modeling projects from complex data without the use of additional feature engineering approach on dataset?

Hello, I have created a roadkill prediction model using MaxEnt and have inputted the .asc output into ArcPro. I would now like to perform further analysis using ArcPro 'Optimized hot spot analysis' tool. I have seen this can be done in multiple scientific papers.

I have tried to convert my .asc file into points using 'Raster to points' tool. Unfortunately, this did not work.

Does anyone know a way to transform the .asc file into point data or perform hot spot analysis on an .asc file without transformation.

Thank you for your time!

I'm looking for long term load histories from real structures to training some AI prediction models. These must be real signals, not a simulation result. The measured quantities can be accelerations (e.g. piezoelectric sensors) or even better strains (e.g. strain gauges). By long-term runs, I mean a length of several months to several years, or many recordings of similar load cycles (e.g. an airplane flight cycle or a typical excavator working day). Maybe this will be the beginning of an interesting collaboration on a scientific article? Anyone know, anyone seen :-)

Hello every body,

I need references that talk about predictive models of corrosion (uniform and localized) of oil and gas metal storage tanks.

Dear Colleagues I hope all is well. I have a classification project, the event outcome is imbalanced (25% of the subjects having the event) The data has almost 90 variables that become 250 after doing one hot encoding I tried oversampling technique and the accuracy, sensitivity are really very well in the oversampled training data (just excellent, with accuracy up to 95%). However, it is very bad, on the validation set (almost 25%). This happens when using Logistic regression, random Forest, decision tree, XGBOOST, gradient boost and Bagging. I wonder if this would be related to the big number of features (250 features?) Shall I do Recursive feature elimination Random Forest before running all these models on the oversampled data? Would this make a difference? Or Recursive feature elimination Random Forest is only used at the end to get a simplified prediction model?

Given a business problem, there is no hard and fast rule to determine the exact number of neurons and hidden layers required to build a neural network architecture. The optimal size of the hidden layer in a neural network lies between the size of the output layers and the size of the input. However, here are some common approaches that have the advantage of making a great start to building a neural network architecture –

To address any specific real-world predictive modeling problem, the best way is to start with rough systematic experimentation and find out what would work best for any given dataset based on prior experience working with neural networks on similar real-world problems. Based on the understanding of any given problem domain and one’s experience working with neural networks, one can choose the network configuration. The number of layers and neurons used on similar problems is always a great way to start testing the configuration of a neural network.

It is always advisable, to begin with, simple neural network architecture and then go on to enhance the complexity of the neural network.

Try working with varying depths of networks and configure deep neural networks only for challenging predictive modeling problems where depth can be beneficial.

I am working on predicting some parameters in soil using visible-near infrared reflectance data. My question is 'What is the minimum number of soil samples and their corresponding reflectance data required for generating a prediction model using ANN machine learning algorithm?'.

for example, how can I calculate c-index or c-statistic (for investigation of my prediction model performance) at 5-year and 10-year follow-up in cohort sample?

I am using photo-fenton catalyst to degrade different types of Persistent Organic Pollutants. Upon degradation we have analysed the degraded product using GC-MS and LCMS. To further analyse the toxicological effect of degraded product we want to do ecotoxological analysis using ECOSAR predictive model. We are seeking collaboration who can help us on this.

I've built a data matrix that contains 7 independent variables. When I performed linear regression using this matrix, I noticed that one of the independent variables is displayed as an exponential function, whereas the rest is plotted linearly.

I'm aware (although not experienced) that it is possible to perform non-linear regression on this single non-linear independent variable, but I'm not sure how to combine the data of the linear and non-linear regression into a single predictive model.

I would be very grateful if anyone could share some insights into this problem!

I want to build an LSTM-based model (where all the outputs are the same features as the input, four inputs, and four outputs) to forecast the next measurement of a sensor node. The scenario is: after receiving the first input (a multivariate data point), the model predicts the second, then gets the second input and predicts the third based on the first and the second values, and so on.

I am aiming to assess habitat suitability for a fish species. I would like to understand multi-scale habitat variables and be able to create a predictive model/map from this data. It seems like Random Forest modeling would be a great application for this, however, I would like to account for imperfect detection. I have seen that random forest is better at dealing with non-linear effects compared to occupancy models. I have also read that imperfect detection can really obfuscate the predictions from a Random Forest model.

I have seen a few groups using them in 'conjunction' (comparing predicted results from RF and Occupancy models) but it never seems like method actually compliments the other. I could be wrong.

Any insight into clear advantages of one over the other for my application would be greatly appreciated. Additionally, if anyone has seen both used in tandem, in a sensible manner that would also be great.

Thanks so much!

Hello Experts,

Since our targeted species is found only in the 2 km region of the study site, we are planning to use 30 m spatial resolution climate data on our Species Distribution Model. But the problem is that my local weather station is capable of providing 20 km resolution data. On the other hand, if I use WorldClim data that is also 1 km.

My questions are

1. Can I use these downscaled data (from 1 km or 20 km) on my local study on SDM, which will be on 30 m resolution?

Thank you for your valuable time.

Hello Experts,

My study site is relatively small and the targeted species is found as continuous patches. Do I need to consider Patch size/area in the MaxEnt model?

Does patch size have any meaningful measurable values that can be included in the MaxEnt model?

Thank you.

Hello Experts,

We are at the beginning of making predictive modelling of an invasive plant species using MaxEnt. The species is found as a patch over the study area. I am new at using this model, have a piece of limited knowledge about it. I have reviewed several papers where only point locations of the present occurrence had been used.

Since my target species occurs as a patch, How can I take the polygonal area of the species where it occurs, instead of point location data?

Or are there any other methods to cover the whole patch of the species into SDM?

Dear colleagues.

I am thinking about the possibility of developing a predictive model applied to technology transfer processes in healthcare, its main intention in anticipating the success achieved by the parties involved. Computationally, is this possible?

Predictive models that use ordinary least squares (OLS) for parameter estimation must show residuals with normal distribution and constant variance (homoscedastic).

However, in most scientific articles (in engineering-related areas, at least) I don't see a concern with meeting these assumptions. In your opinion, why does this happen? In the end, the results do not change that much when we make the necessary transformations so that these assumptions are met?

If you have had any experience with this topic, please feel free to share.

Respected all,

I am using regression modelling for crash prediction model. Can anyone tell me what should be the minimum value of Chi square for accepting in Publication?

Thank you all in advance.

A case-control study has been used to construct a prediction model and there is no cohort study to calibrate it.

What solution do you suggest for calibration?

While working in both the software, after loading the training and validation data for the prediction of a single output using several input variables (say 10), it skips some of the inputs and delivered an explicit mathematical equation for future prediction of the specific parameter but it skips some of the input variables (say 2 or 3 or maybe greater). What criteria are these software uses in the back for picking the most influential parameters while providing a mathematical predictive model?

I have developed a logistic regression based prognostic model in Stata. Is there any way to develop an app using this logistic regression equation (from Stata)?

Most of the resources I found require me to develop the model from scratch in Python/R and then develop the app using streamlit/Shiny etc.

However, I am looking for a resource where I could use the coefficients and intercept values from Stata based model rather than build model from scratch in python.

Dear researchers

When the data set for a particular prediction is unbalanced( negative class is more than positive class OR negative class is less than positive class), Which evaluation parameter is appropriate to evaluate the performance of the predictive model? Why?

These parameters are including Specificity(SPE), Sensitivity(SEN), Receiver operating characteristics (ROC), Area Under Curve (AUC or AUROC), Accuracy(ACC), and f-measure(F-M)

Thanks for your guidance

I am waiting for your answer

I have the following dataset:

1 - male - 15KG - abc - Year 1 - 2 - quantity 1

- Year 2 - 3 - quantity 2

2 female - 17KG - cde - Year X - 4 - quantityx

- 16KG - Year Y - 6 - quantityy

- Year Z - 3 - quantityz

.... etc...

I want to make a prediction model that predict the quantity, but using classic machine learning models ( not deep learning ones, like LSTM or RNN ), i.e. linear regression, SVM , .. such that:

predict quantity of n individual at a certain duration ( duration A) what will be the quantity ?

n - male - 25KG - xlm - 34 - A - ?

What is the best was to treat and pre-process duration , trail and quantity features before fitting them to preserve their correlation with the target quantity ?

I build an predictive machine learning model that generate the probability to default over the next 2 years for all the companies in a specific country. I used for training the algorithms financial data for all these companies and also the NACE codes (domains of activity) and I'm wondering if I will develop a better model if I somehow segment the population of B2B in segments and run distinct models on these segments.

Hope you can advice!

Lots of thanks in advance!

I read some papers of risk prediction models, and generally there are three types of measures of the performance: a) Total model performance (e.g. R^2); b) Discrimination; c) Calibration.

However, many papers of risk prediction model only reported the Discrimination and Calibration, seldom reported the R^2. What confused me is that if R^2 is important in risk prediction models. And is there a threshold for the minimal value of R^2 in such models?

I previously installed minitab software. Tried to use it on non-linear regression but having trouble to find good starting values during the curve fitting process.

(primary models; Gompertz, Logistic, Weibull, Baranyi)

The goal of predictive analysis is to develop predictions for the development of complex, multifaceted processes in various fields of science, industry, economy or other spheres of human activity. In addition, predictive analysis may refer to objectively performing processes such as natural phenomena, climate change, geological, cosmic etc.

Predictive analysis should be based on taking into account in the analytical methodology possible the most modern prognostic models and a large amount of data necessary to perform the most accurate predictive analysis. In this way, the result of the prediction analysis performed will be the least subject to the risk of analytical error, ie an incorrectly designed forecast.

Predictive analysis can be improved by using computerized modern information technologies, which include computing in the cloud of large data sets stored in Big Data database systems. In the predictive analysis, Business Intelligence analytics and other innovative information technologies typical of the current fourth technological revolution, known as Industry 4.0, can also be used.

The current technological revolution known as Industry 4.0 is motivated by the development of the following factors:

Big Data database technologies, cloud computing, machine learning, Internet of Things, artificial intelligence, Business Intelligence and other advanced data mining technologies. On the basis of the development of the new technological solutions in recent years, dynamically developing processes of innovatively organized analyzes of large information sets stored in Big Data database systems and computing cloud computing for the needs of applications in such areas as: machine learning, Internet of Things, artificial intelligence, Business Intelligence are dynamically developing.

For the abovementioned application examples, one can add predictive analyzes of subsequent, other fields of application of advanced technologies for the analysis of large data sets such as Medical Intelligence, Life Science, Green Energy, etc. Processing and multi-criteria analysis of large data sets in Big Data database systems is carried out according to V4 concepts, ie Volume (meaning a large number of data), Value (large values of certain parameters of the analyzed information), Velocity (high speed of new information) and Variety (high variety of information).

The advanced information processing and analysis technologies mentioned above are used more and more often for the needs of conducting predictive analyzes concerning, for example, marketing activities of various business entities that advertise their offer on the Internet or analyze the needs in this area reported by other entities, including companies, corporations, institutions financial and public. More and more commercial business entities and financial institutions conduct marketing activities on the Internet, including on social media portals.

More and more public institutions and business entities, including companies, banks and other entities, need to conduct multi-criteria analyzes on large data sets downloaded from the Internet describing the markets on which they operate, as well as contractors and clients with whom they cooperate.

On the other hand, there are already specialized technology companies that offer this type of analytical services, including offering predictive analysis services, develop custom reports, which are the result of multicriteria analyzes of large data sets obtained from various websites and from entries and comments. contained on social media portals based on sentiment analyzes of the content of entries in the comments of Internet users.

Do you agree with my opinion on this matter?

In view of the above, I am asking you the following question:

Please reply

I invite you to discussion and scientific cooperation

Dear Colleagues and Friends from RG

I invite you to discussion and cooperation.

Best wishes

We are trying to calibrate/validate our predictive productivity/erosion model (APEX) against multi-year productivity data for Kenya and Namibia. Erosion measurements also needed. FAO datasets do not provide raw values, so are not good for us. We are looking for ground measurements in any location throughout either Kenya or Namibia. Can anyone help? Thanks.

We are trying to calibrate/validate our predictive productivity/erosion model (APEX) against multi-year productivity data for Kenya and Namibia. Erosion measurements also needed. FAO datasets do not provide raw values, therefore they are not good for us. We are looking for ground measurements in any location throughout either Kenya or Namibia. Can anyone help? Thanks.

How to introduce a covariate in a multivariable logistic prediction model when the event of interest always occurs in individuals exposed to this covariate. It is possible to determine the adjusted odds ratio with their 95% confidence interval?

I almost asked this as a "technical question" but I think it's more of a discussion topic. Let me describe where I get lost in this discussion, and what I'm seeing in practice. For context, I'm a quantitative social scientist, not "real statistician" nor a data scientist per se. I know how to run and interpret model results, and a little statistical theory, but I wouldn't call myself anything more than an applied statistician. So take that into account as you read. The differences between "prediction-based" modeling goals and "inference-based" modeling goals are just starting to crystalize for me, and I think my background is more from the "inference school" (though I wouldn't have thought to call it that until recently). By that I mean, I'm used to doing theoretically-derived regression models that include terms that can be substantively interpreted. We're interested in the regression coefficient or odds ratio more than the overall fit of the model. We want the results to make sense with respect to theory and hypotheses, and provide insight into the data generating (i.e., social/psychological/operational) process. Maybe this is a false dichotomy for some folks, but it's one I've seen in data science intro materials.

The scenario: This has happened to me a few times in the last few years. We're planning a regression analysis project and a younger, sharper, data-science-trained statistician or researcher suggests that we set aside 20% (or some fraction like that) of the full sample as test sample, develop the model on our test sample, and then validate the model on the remaining 80% (validation).

Why I don't get this (or at least struggle with it): My first struggle point is a conceptual/theoretical one. If you use a random subset of your data, shouldn't you get the same results on that data as you would with the whole data (in expectation) for the same reason you would with a random sample from anything? By that I mean, you'd have larger variances and some "significant" results won't be significant due to sample size of course, but shouldn't any "point estimates" (e.g., regression coefficients) be the same since it's a random subset? In other words, shouldn't we see all the same relationships between variables (ignoring significance)? If the modeling is using significance as input to model steps (e.g., decision trees), that could certainly lead to a different final model. But if you're just running a basic regression, why would anyone do this?

There are also some times when a test sample just isn't practical (i.e., a data set of 200 cases). And sometimes it's impractical because there just isn't time to do it. Let's set those aside for the discussion.

Despite my struggles, there are some scenarios where the "test sample" approach makes sense to me. On a recent project we were developing relatively complex models, including machine learning models, and our goal was best prediction across methods. We wanted to choose which model predicted the outcome best. So we used the "test and validate" approach. But I've never used it on a theory/problem-driven study where we're interested in testing hypotheses and interpreting effect sizes (even when I've had tens of thousands of cases in my data file). It always just seems like a step that gets in the way. FWIW, I've been discussing this technique in terms of data science, but I first learned about it when learning factor analysis and latent variable models. The commonality is how "model-heavy" these methods are relative to other kinds of statistical analysis.

So...am I missing something? Being naive? Just old-fashioned?

If I could phrase this as a question, it's "Why should I use test and validation samples in my regression analyses? And please answer with more than 'you might get different results on the two samples' " :)

Thanks! Looking forward to your insights and perspective. Open to enlightenment! :)

Hello everyone, I am currently working on a predictive maintenance model to detect possible failures in a specific system or equipment. I wanted to ask you if you know where I can get real data from a device from a normal operating state to a failed state, such as valves, motors, switches, etc.