Credit and Finance Risk Predictive Modeling & Management

Credit risk analysis (finance risk analysis, loan default risk analysis) and credit risk management is important to financial institutions which provide loans to businesses and individuals. Credit can occur for various reasons: bank mortgages (or home loans), motor vehicle purchase finances, credit card purchases, installment purchases, and so on. Credit loans and finances have risk of being defaulted or delinquent. To understand risk levels of credit users, credit providers normally collect vast amount of information on borrowers. Statistical predictive analytic techniques can be used to analyze or to determine risk levels involved on credits, finances, and loans, i.e., default risk levels. Credit risk predictive modeling is discussed here.

Why internal credit scoring?

Personal credit scores are normally computed from information available in credit reports collected by external credit bureaus and ratings agencies. Credit scores may indicate personal financial history and current situation. However, it does not tell you exactly what constitutes a "good" score from a "bad" score. More specifically, it does not tell you the level of risk for the lending you may be considering. Furthermore, in many countries, credit rating system is not available. Internal credit scoring methods described in this page address the problem. It is noted that internal credit scoring techniques can be applied to commercial credits as well.

Credit Risk Analysis and Modeling

In this page, the following credit risk analysis methods are described;

  • Credit risk factors (hotspot) profiling or loans default analysis.
  • Credit risk predictive modeling or loans default predictive modeling.
  • Credit risk modeling or finance risk modeling.
  • Internal credit risk scoring.

Hotspot Profiling of Risky Credit Segments

Credit risk profiling (finance risk profiling) is very important. The Pareto principle suggests that 80%~90% of the credit defaults may come from 10%~20% of the lending segments. Profiling the segments can reveal useful information for credit risk management. Credit providers often collect a vast amount of information on credit users. Information on credit users (or borrowers) often consists of dozens or even hundreds of variables, involving both categorical and numerical data with noisy information. Hotspot profiling is to identify factors or variables that best summarize the segments.

CMSR Data Miner hotspot profiling tools drill-down data systematically and detect important relationships, co-factors, interactions, dependencies and associations amongst many variables and values accurately using Artificial Intelligence techniques, and generate profiles of most interesting segments. Hotspot analysis can identify profiles of high (and low) risk loans accurately through thorough systematic analysis of all available data.


Credit Risk Predictive Modeling

If past is any guide for predicting future events, predictive modeling is an excellent technique for credit risk management. Predictive models are developed from past historical records of credit loans, containing financial, demographic, psychographic, geographic information, etc. From the past credit information, predictive models can learn patterns of different credit default/delinquency ratios, and can be used to predict risk levels of future credit loans. It is important to note that statistical process requires a substantially large number of past historical records (or customer loans) containing useful information. Useful information is something that can be a factor that differentially affects credit default/delinquency ratios.

Predictive Models on Android Devices (Downloads)

If you are interested to know how predictive models can be used to manage risk, try MyDataSay Android App. Download is available from the linked page.

Credit Risk Predictive Modeling Methods and Software Tools

CMSR Data Miner supports robust easy-to-use predictive modeling tools. Users can develop models with the help of intuitive model visualization tools. CMSR supports the following predictive modeling tools;

  • Neural Network is a very powerful modeling tool. It generally offers most accurate and versatile models. It's very easy to develop neural network predictive models with CMSR. Network visualization tools will guide users from configuration, training, testing, and more importantly direct application to databases.
  • Cramer Decision Tree produces most compact and thus most general decision trees. Decision tree can be used for predicting segmentation-based statistical probability of credit loan defaults.
  • Regression produces mathematical functions for predicting default risk levels. It can be very limiting to be used as general-purpose credit risk predictive modeling methods. However when it is used with above methods, it can be a very useful method.
  • RME (Rule-based Model Evaluation) is a powerful model integration tool. It can be used to combine a number of predictive models into a single model, producing combined predictions such as maximum, minimum, average, etc. In addition, it can be used to classify combined predictions into classes such as "Very high risk", "High risk", "Medium risk", "Low risk", etc.
  • RME-EP (Rule-based Model Evaluation with Event Processing) is an amalgam of predictive modeling and forward chaining rule engine. It provides a powerful platform for Deep Learning style modeling. Very intelligent models can be developed.

Does Predictive Modeling Work?

Effectiveness of predictive modeling depends on the quality of historical data. If historical data contains information that can predict customer tendencies and behaviors, predictive modeling can be very effective. Otherwise reliable predictive models will be difficult to obtain. How can you know whether your customer data contain predictive information? You need to perform variable relevancy analysis and build models and test!

Do you have recent historical data?

If your organization has data that can be used to develop risk predictive models, please try CMSR Data Miner. Apply download from CMSR Download Request. One year extended free trial program is available. You will also receive "Predictive Modeling Guide to Credit and Insurance Risk Scoring" ebook and modeling tips.

If you have questions regarding predictive modeling described here, please write to us.


Credit Risk Scoring

Credit risk score is a risk rating of credit loans. It measures the level of risk of being defaulted/delinquent. The level of default/delinquency risk can be best predicted with predictive modeling. Credit risk scores can be measured in terms of default/delinquency probability and/or relative numerical ratings. The following subsections outline credit risk scoring methods;

Why Neural Network and Deep Learning?

A commonly used method used in risk prediction is regression. Regression works well if information structure is functional and simple. However it does not perform well on complex information with many categorical variables. Another commonly used method is decision tree. Decision tree is not suitable if dependent variables have heavy skews. Credit loan data have this skew. This leads neural network and deep learning methods to be the choice for credit risk modeling. The following figure shows a neural network model;

Neural Network

Neural network arranges information in nodes and weight-links as shown in the above figure. Nodes represent input/output values. Nodes are organized into layers: input layer, (optional) internal layers (normally a single layer as in the figure), and output layer. Input layer nodes accept input values. Values of output layer nodes and internal layer nodes are computed by summing up previous layer nodes multiplied by weight-links' values.

Neural network weight-links are computed in such a way that given input values, network produces certain output value(s) for output layer node(s). This process is called as network training. This is performed using past data. Neural network is a heuristic predictive system.

Bias nodes are similar to coefficients in regression. They have value 1 and tend to improve network's learning capability.

In the above chart, positive value weight-links are colored in red. Negative value weight-links are colored in blue. Colors are scaled according to absolute value ratios against the largest absolute value. Absolute value zero is colored in black. Largest absolute value is colored in pure red or blue color. The rest are scaled accordingly.

It is noted that neural network is not good at predicting unseen information. It can make very wild predictions. Thus good training data is very important.

In the following sections, credit risk modeling steps are described.

Step 1: Develop Neural Network Models

Predictive models infer predictions from a set of variables called independent variables. To develop models, the first step is to analyze which variables contain predictive information through relevancy analysis. Once relevant variables are identified, (neural network) models can be configured and trained using past historical data. Neural network training is a repetitive process which may take long. Fast computer may be needed. Fully trained models should be tested using past historical data before using them. Single models can have bias and weakness. To overcome this, multiple models can be developed and combined as described in the next section.

Step 2: Combine Neural Network Models

Once models are fully trained and tested, they can be integrated to produce combined outputs such as largest (=maximum), smallest (=minimum), average, average without largest and smallest values, etc. This can be done using RME/RME-EP (Rule-based Model Evaluation available in CMSR Data Miner) easily.

The following histogram shows largest(=maximum) scores and risk distribution in past historical data. "MaximumScore" represents the combined largest(=maximum) values horizontally. Vertically risk (=delinquent/defaulted) proportion is shown. (Note that the label "Risky" represents historical data which were in delinquent/default state. It is used because it makes sense in classification modeling.) It clearly shows that higher scores have higher proportion of risk in the past historical data. So the models are effective and useful. Note that the neural network models are trained to predict values between 0 and 1. This can be a bit higher and a bit lower value as seen in the histogram.

Histogram of maximum risk scores

The following figure plots data of the above chart. "RID" record identifier is used to spread data horizontally so that data can be seen easily. Vertically it shows maximum values scored by models. Red circles represent historical data records that were in fact delinquent or defaulted. Clearly this plot shows higher the score is, higher the risk. Score 0.6 and above was all delinquent/defaulted. Score 0.4 to 0.6 also has high risk. Score 0.2 to 0.4 has medium risk. The rest has low risk.

Scatterplot of maximum risk scores

Deep Learning and Artificial Risk Management Expert Advisor

(Good news! You can develop Risk Management Expert Advisor Models using Deep Learning techniques very easily. What you need is relevant data for Deep Learning.)

Large neural networks are very powerful. They can learn finely detailed information. This will lead to overfitting. This is no good for predictive modeling which requires statistical learning. To overcome this overfitting problem, Deep Learning style modeling can be used. Instead of large networks with large number of input and hidden layer nodes, multiple smaller specialized neural networks performing smaller decomposed tasks can be used. In addition, regression and decision tree models can be incorporated into deep learning models. Multiple smaller neural networks are then integrated with integration neural networks. Advanced deep learning models can work as Artificial Risk Management Expert Advisor. For more on this, read Deep Learning and Rule Engine with Predictive Modeling.

Step 3: Risk Scores to Risk Classification

Risk scores produced by neural network and RME/RME-EP models can be confusing to users. It will be better if they are verbalized into more easily understood vocabularies such as "Very high risk", "High risk", "Medium risk", "Low risk", etc. The above histogram clearly shows that if maximum risk score is equal or greater than 0.6, it has 100% risk. So it can be coded as "Very high risk". The next class is if maximum risk score is equal greater than 0.3, it has "High risk". The next class is if maximum risk score is equal greater than 0.2, it has "Medium risk". The rest has "Low risk". This classification produces risk distribution as in the following chart.

Proportions of risk classes

This chart shows how each class had risk in the past historical data. This classification is coded using an RME/RME-EP model. You need two RME/RME-EP models: One is to combine scores to produce maximum scores for analysis. The next model is to produce classification and to deploy.

This classification can be extended to include minimum and average risk scores. Expanded documentation of this extension can be found in MyDataSay Android Application. Download is available here MyDataSay Android App.

* Note that models and charts used in this page are produced using CMSR Data Miner and are based on artificially generated data.

Step 4: Deploy Models for Users

Once models are fully trained, tested and combined into RME/RME-EP models, they are ready to deploy for customer-facing users. We provide the following deployment options;

  • CMSR Data Miner: This provides model development and batch database updates. Models can be applied to database tables directly from CMSR Data Miner.
  • Web Server: Rosella BI Server provides predictive models to users through web. It can support a large number of users. It is optimized for small screen devices such as smart phones and tablets as well as normal computers. In addition, it can be incorporated into your web-based business applications using Java JSP pages and JSON HTTP requests. For more, read Rosella BI Server.
  • Android Application: MyDataSay is an Android application which can be used by an unlimited number of users. Download is available here MyDataSay Android App. You are recommended to download and try MyDataSay. You can learn how predictive modeling can be used.

The following figure shows web deployment of predictive models. (It is important to note that your local anti-discrimination laws may bar using certain variables in your predictive models, such as gender, race, age, etc. if models are used to discriminate customers. Check your local laws.)
predictive model deployment.

For more details on these modeling steps, please read Nine Steps Predictive Modeling Guide for Risk Management and Predictive Modeling Cook Book.

Predictive Models on Android Devices (Downloads)

If you are interested to know how predictive models can be used to manage risk, try MyDataSay Android App. Download is available from the linked page.

For information about predictive modeling, please read Predictive Modeling Software Tools.