Customer Churn

Focus Area: Classification Model

Project: Predict likeliness of a given customer to churn

Tools: Random Forest, SMOTE

Read the full writeup on GitHub

Make it stand out.

How often do you change phone carriers? This project aimed to create a model that could reliably predict whether or not a given customer would cease using their telecommunications company's services.

I used past customer churn data from Telco. Similar models can be created to fit other industries and their nuances.

Once you understand the top reasons for churn, you can dial in on how to most efficiently retain customers and market to ideal new ones.

Here I found that those not specifically subscribed to tech support or online security [regardless of how many other upsell-programs they are subscribed to], are more likely to churn.

These are two of the most paramount factors; highly recommended for telecoms to focus marketing efforts on.