Project Overview

This project focuses on predicting whether a user would stop using a particular service or not. We perform EDA on the data and proceed to test different models on it to see which gives us the best accuracy.

We start off by doing some Exploratory Data Analysis of our data. We check for any missing or erroneous values.

Correlation average call and email replies


We also look at how each feature correlates to the others but most importantly, to the target variable. It is important to note that we also have to check if any features are too highly correlated so that we can drop them to avoid multicolinearity.

Months customer has been using service


Next we also add some new features that would help us train our model and hopefully extract some value from these new features.

Finally, we apply different models to our data and see how they do. Once we find the models with the best accuracy, we use SHAP to identify how each model makes the predictions that it does.

Head on over to the link below to check out the entire python notebook which contains the entire step by step process of first carryinh out the EDA, then we move on to the training of different models on our data and lastly explore the SHAP values of the best models.

• Repository