Predicting insurance company customer renewal probability and optimizing sales agents time spent on each customer.
Solution for McKinsey Data Scientist Hiring Hackathon. https://datahack.analyticsvidhya.com/contest/mckinsey-analytics-online-hackathon-4/
Techniques used:
Light Gradient Boosting Model (LGBM) to build a high accuracy tree model.
RandomizedSearchCV (Scikit learn) to train LGBM with different combinations of hyperparameters using CrossValidation.
Scipy to find input value that maximizes outoput for a function.