Photo by Michael Jasmund on Unsplash
This project is a ML approach to predict customer churn.
Predicting the churn allows the company to get ahead of when a client might leave and to react in time to prevent it from actually happening. This can be done with special offers for example, trying to change his mind about leaving and it's only possible because of previous data about that particular client as opposed to a new client the company knows nothing about yet.
Retaining customers can be easier than attracting new ones since these are likely less interested in what you're offering and there's no connection between you yet so you would be starting from scratch. In addition to that retaining a client will cost the company less than attracting new ones.
The entire process is detailed and coded in the notebooks: PT or EN
Also, a blog post of the findings is available here.
Kaggle for providing the Telecom Users Dataset.