I build predictive models about policy renewal using Logistic Regression and Random Forest.
If a company interacts with every single customer, the cost would be too much. Focusing retention efforts on a small subset of high risk policies is a much more effective strategy.
In this post, I’m going to focus on Random Forest as a classification method. Unlike logistic regression, random forest is better at fitting non-linear data. It can also work well even if there are correlated features, which can be a problem for interpreting logistic regression.
Random forest uses multiple decision trees to make predictions. Single decision trees on their own can be very effective at learning non-linear relationships (low bias, but high variance). Here is an exemple of a simple tree with 3 only features:
Due to their high variance, they can tend to over-fit. Random forest reduces this variance by averaging many trees (at the sacrifice of a slight increase in the bias).
In this case, I fit the model to 14 features. So by default, randomForest() R function will set mtry = sqrt(14) = 3.74 rounded down = 3. Also, by default random forest generates 500 trees, but 200 would be enough as errors rate do not decrease as we can see in next plot:
After I fit the model, let’s take a look at the confusion matrix to see how well the model made predictions on the validation set. The accuracy is pretty good!
And in the next plot, we can study the importance of each feature in the prediction model:
