Audiobook churn prediction

This project involves forecasting whether an audiobook user will be able to retain or churn from a vendor company. Data comes from an unspecified audiobook vendor. This repository contains a well-commented jupyter notebook that walks along the path of prediction process. This involves 5 stepe:

1. Introduction
2. Data Exploration
3. Data Cleaning and Feature Engineering
4. Model Building
5. Results and Discussion

In the modelling phase a number of classic and modern models are run and explored. The tree based ensemble models seem to perform better in terms of model performance metrics.