Marketing Analytics with R - Customer Purchasing Behavior Analysis

Research Questions In the field of marketing, one problem businesses try to address is: how customer demographics and preferences influence customer purchasing behavior? This is important. In my case, I am working on analyzing car insurance company data. The 3 questions I want to address are: 1. Who will accept the Car Insurance Company’s offers? 2. What characteristics do these customers have? 3. Who should this Car Insurance Company target on? Analysis To begin with, I have made the goal clear. The project is about customer targeting. I faced several problems: * Highly imbalanced dataset * Model choice * Variable selection I solved these problems by: * Using several sampling methods, including under-sampling method, random over-sampling method and SMOTE+ENN, to generate a reasonably balanced dataset. I utilized AUC as metric to choose proper sampling method and we chose SMOTE+ENN. * I used logistic regression model to build model because the result of logistic regression model is easy to interpret. As my goal is to find which customer to target, I need the model result to be interpretable. * I tried dummy variables, interaction variables, non-linear transformation to select proper variables. The detailed R code can be found in this link.