Using techniques developed in the area of Survival Analysis (usually used for drug trials) to predict Telco customer churn.
This is actually a pretty useful idea as customers churning is anologous to patients dying.
- pandas
- numpy
- tqdm (optional)
- matplotlib (optional)
- lifelines
- sklearn
- This code is my attempt to get a foot hold in this area. As such it is basically a copy of Carl Dawson's Churn Prediction and Prevention artical. All credit for the creation of this code should go to him and his sources.
- It should also be noted that in the real world data is messy, so the data used here, avaliable via Kaggle's Telco Customer Churn website, is only an idealised example. From expereince, data, especially Telco or large company data is messy and a large portion of the development for any project like this is involved in getting good data, especially if it needs to be labelled.