Project Description

 

Predicting customer app behavior in terms of purchase a product using Machine Learning

The model is predicting if a user, that can enjoy a 24 hours free trial of the premium features of the app, is going to purchase the product or not after the free trial is over.

Some of the features used to build the model are: when and how the user has been navigating into the app, this information has been compiled during the trial period, and is what we are using to predict how likely the user is going to pay for the app. We could target them who are more ”unlikely” be enrolled with a customized offer, focusing the marketing efforts into them.

Finally the Logistic Regression model gives 70-76% accuracy in pur predictions, this percentage could increase tune hyperparameters and EDA, also if we run our predictions on dailiy basis the model could get more consistent.

It infers from the results that the more late the day of the week , the younger the user us and the more number of screen the user see, the more engaged with the app the user is, and more likely to be enrolled the user is.