Refunds reduction
Tom Papantos
Data Scientist
Marketing Analytics Specialist
Statistician
Python
scikit-learn
Visual Studio Code
Mott & Bow
Refunds is a huge pain point for businesses, therefore we wanted to understand which variables reduced the probability of a refund happening (same approach as used while maximizing customer LTV project).
Steps applied:
Took the dataset at an item level
Ran a logistic regression.
Coefficients interpretation.