Build Churn Model
Summary: The goal of this study is
- Identify and visualize the factors that lead customers to terminate their credit cards.
- Build a prediction model that used for classifying if a customer is going to churn or not.
Methodology: Firstly, I do Exploratory Data Analysis with Visualization based on Demographic data and Banking Profile Secondly, I do Feature Engineering to rescale numeric data and form new features based on existing features. Thirdly, I do Data Augmentation to deal with imbalance data.ΒΆ Finally, I train 5 models (Logistic Regression, Support Vector Classifer, KNN, Random Forest and XGB) and do Model Evaluation to choose the best ones to apply in another test data set.