Churn_Project

In this project, I performed several crucial data preprocessing steps, including One-Hot Encoding (OHE) for handling categorical variables, scaling the numerical features to standardize the dataset, and addressing outlier detection and handling to ensure data quality. To deal with the class imbalance in the dataset, I applied SMOTE (Synthetic Minority Over-sampling Technique) to balance the classes. After preprocessing, I split the data into training and validation sets, and built a Random Forest Classifier with balanced class weights. I fine-tuned key hyperparameters like the maximum depth of the trees, the number of estimators, and the minimum number of samples required for splitting or creating leaf nodes. I evaluated the model’s performance using accuracy, confusion matrix, and classification report on the validation set. Finally, I generated predictions on the test dataset, combined them with customer IDs, and saved them to a CSV file for autograding, ensuring that it met all required specifications.