Applied Churn Modelling Problem using ANN-Artificial Neural Network and Random Forest Classifier models for predicting chances of a customer leaving telecom service based on 10+ parameters, solved the problem of imbalanced dataset by using SMOTE and improved the f1 score from 58% to 92% and accuracy from 88% to 92%.
sudhanshublaze / churn_prediction Goto Github PK
View Code? Open in Web Editor NEWData set: https://www.kaggle.com/c/customer-churn-prediction-2020/
License: Apache License 2.0