Process

Repurchasing Behavior

Purpose

We use statistical learning methods, supervised and unsupervised classification analytics, to build our customer repurchasing model for the online insurance business.

Discuss issues

Issue 1. We want to know the repurchasing time period between the first-time and the second-time subscribing and possible factors.

Issue 2. From the repurchasing model we build, we want to explore reasons behind customer repurchasing behavior.

Targeted customer

For the first-time subscriber of online traveling insurance, we use the repurchasing model to identify the possibility of repurchasing.

Expected achievement

Use the repurchasing model to identify the potential customers and all those identifications from the model are written as selective rules or conditions.

The procedures of execution

Step 1. To integrate different customer data sets, both new customers with old customer in the old database are included. We may need to perform the missing value replacement or data cleansing to have a higher quality coherent dataset.

Step 2. With data mining results, we define new categories for further analytics. We have to design some new variables for the time period between two purchases to explore important factors on the customer repurchasing behavior.

Step 3. Perform the decision-tree statistical learning method for specific repurchasing issues to build the repurchasing model with clear rules of classifications. Decision-tree statistical learning method is used on the integrated high quality data set to have classification rules we could follow with further marketing strategy.

Step 4. With the combination of unsupervised clustering method and supervised decision tree learning, we rebuild a customer purchasing model on specific groups to explore the possibility of repurchasing behavior. The classification rules and conditions are provided from new repurchase modeling.