|Rosella Machine Intelligence & Data Mining|
Insurance Risk Prediction Modeling and Insurance Risk Analysis
Risk management is very important for insurance industry. Insurance means that insurance companies take over risks from customers. Insurers consider every available quantifiable factors to develop profiles of high and low insurance risk. Level of risk determines insurance premiums. Generally, insurance policies involving factors with greater risk of claims are charged at a higher rate. With much information at hand, insurers can evaluate risk of insurance policies at much higher accuracy. To this end, insurers collect a vast amount of information about policy holders and insured objects. Statistical methods and tools based on data mining techniques can be used to analyze or to determine insurance policy risk levels. Insurance risk predictive modeling is discussed here.
Insurance Risk Analysis and Predicting Insurance Claims RiskIn this page, insurance risk analysis and methods for predicting insurance claims prediction are described;
Hotspot Profiling of Risky Insurance Segments
Profiling insurance risk factors is very important. The Pareto principle suggests that 80%~90% of the insurance claims may come from 10%~20% of the insurance segment groups. Profiling these hotspot segments can reveal invaluable information for insurance risk management. Insurance providers often collect a large amount of information on insured entities. Policy information (such as automobile insurance, life insurance, general insurance, etc.) often consists of dozens or even hundreds of variables, involving both categorical and numerical data with noisy information. Profiling is to identify factors and variables that best summarize the segments.
CMSR Data Miner / Machine Learning / Rule Engine Studio 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 such as incremental learning and searching, and generate profiles of most interesting segments. It is noted that insurance premiums are normally stipulated with profiles of risky (or very low-risk) policy holders. Hotspot analysis can identify profiles of high (and low) risk policies accurately through thorough analysis of all available insurance data.
For more on customer risk hotspot profiling, please read customer profiling.
Insurance Claims Prediction/Predictive Models by Machine Learning
If past is any guide for predicting future events, predictive modeling by Machine Learning is an excellent technique for insurance risk management. Insurance claims prediction models are developed from past historical records of insurance polices, containing financial, demographic, psychographic, geographic information, along with properties of insured objects. From the past insurance policy information, insurance claims predictive models can learn patterns of different insurance claim ratios, and can be used to predict risk levels of future insurance policies. It is important to note that statistical process requires a substantially large number of past historical records (or insurance policies) containing useful information. Useful information is something that can be a factor that differentially affects insurance claims ratios.
Insurance Claims Predictive Modeling Methods and Software Tools
CMSR Data Miner / Machine Learning / Rule Engine Studio supports robust easy-to-use predictive modeling tools. Users can develop insurance claims prediction models with the help of intuitive model visualization tools. Application and deployment of insurance risk models is also very simple. CMSR supports the following predictive modeling tools;
Does Insurance Claims Risk Predictive Modeling Work?
Effectiveness of insurance claims predictive modeling depends on the quality of historical data. If historical data contains information that can predict customer tendencies and behaviors, insurance claims predictive modeling can be very effective. Otherwise reliable insurance claims risk predictive models will be difficult to obtain. Good historical data is essential to produce good predictive models. How can you know whether your customer data contain predictive information? You need to perform variable relevancy analysis and build models and test!
If you have questions regarding predictive modeling described here, please write to us.
Insurance Claims Risk Scoring by Machine Learning
Insurance claims risk scoring is numerical rating of insurance policies. It measures the level of risk of being claimed. This section describes advanced insurance risk modeling and insurance scoring methods using predictive modeling techniques by Machine Learning;
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. Insurance claims data have this skew. Decision tree can work if claims rate is about, say, 30% ~ 70%. A commonly used method to overcome this problem is the boosting method which dupicates skewed data. But duplication turns outliers into statistically significant patterns, introducing bogus patterns. It also produces lots of false positive predictions. This is a bad approach!
This leads neural network and deep learning methods to be the choice for insurance risk modeling. The following figure shows a neural network model;
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 comprehensive training data is very important. For more on neural network, please read Neural Network.
In the following sections, insurance 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. Only relevant variables are recommended to be used as independent variables. CMSR Data Miner / Machine Learning Studio provides tools for variable relevancy analysis. For more on this, please read Variable Relevancy and Factor Analysis, aka, Principal Component Analysis, for Predictive Modeling.
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-EP (Rule-based Model Evaluation available in CMSR Studio) easily.
Testing predictive insurance risk models requires data visualization tools. The following histogram shows largest(=maximum) scores and risk distribution in past historical data. "MaximumScore" represents the combined largest(=maximum) values horizontally. Vertically risk (=claimed) proportion (in red color) is shown. Note that the label "Risk" represents historical data which were in claimed state. 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.
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 insurance claims. Clearly this plot shows higher the score is, higher the risk. Score 0.6 and above was all insurance claims. Score 0.4 to 0.6 also has high risk. Score 0.2 to 0.4 has medium risk. The rest has low risk.
Step 3: Risk Scores to Risk Classification
Risk scores produced by neural network and 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.
This chart shows how each class had risk in the past historical data. This classification is coded using an RME-EP model. You need two 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 / Machine Learning / Rule Engine Studio and are based on artificially generated data. For software download, please apply CMSR Download Request.
Step 4: Deploy Models for Users
Once models are fully trained, tested and combined into RME-EP models, they are ready to deploy for actuaries and other customer-facing users. We provide the following deployment options;
It is noted that CMSR Studio and Web Server run on Windows, MacOS, Linux, etc.
The following figures show deployment of predictive models. The first figure
is an example of input screen on web browser. The second is an example result on
(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.)
For more details on these modeling steps, please read;