Customer Churn Prediction
This our group's solution to CRM class project to create a customer churn prediction model in order to run an efficient targeted marketing campaign. The final model is a decision tree in order to maintain model interpretability, since this is a consulting case project.
Requirements
- mlr
- dplyr
- tidyr
- ggplot2
- lubridate
- mice
- corrr
- caret
Usage
The scripts need to be executed in the following order:
- preprocessing
- models
- prediction
d18_imputed is required for the models-script to work and model_dt is required for the prediction-script to work.
Details
The preprocessing step takes care of most flaws in the data, including: syntactic & semantic preparations, deduplication, one-hot-encoding, missing value imputation and SMOTe in order to account for the class imbalance.