We are trying to build a model which predicts whether a person purchases an item present in his or her cart on a e-commerce platform.
- Slice and dice data to understand each attribute
- Generate plots to understand the distribution
- Advanced EDA
- Splitting category code at two levels
- Capture User activity (session count, total activity in session, product affinity)
- Binning hours in 4 bucket
- Reduction in brands
- Generate Target variable
- Evaluate these algorithm
- Logistic Regression
- Decision Tree
- Random Forest
- Hyperparameter Tunning (ParamGridBuilder)
- Threshold selection
- Feature Importance
- Model implications