##Customer Churn Prediction – Model Comparison

Hello There! This is a small project I did to predict customer churn — basically, figuring out which customers are likely to leave a service. I worked with the Telco Customer Churn dataset and tried out multiple machine learning models to see which one performs the best.

##Dataset Telco Customer Churn Dataset Target variable: Churn(Yes=1,No=0)

##What I did

1-EDA Checked for missing values Looked at churn distribution Compared churn vs non-churn by tenure and monthly charges

2-Preprocessing Split data into train and test sets Standardized numeric columns One-hot encoded categorical columns

3-Models trained Logistic Regression Random Forest Classifier Gradient Boosting Classifier

4-Evaluation Confusion Matrix Precision, Recall PR-AUC Tried different thresholds for Logistic Regression

5-Feature Importance Checked which features influence churn the most for each model

##Model Comapred Logistic Regression/Random Forest/Gradient Boosting

All those models use identical preprocessing and train/test splits to ensure fair comparison.

Key Results

Gradient Boosting had the best PR-AUC (about 0.66) Decreasing the threshold for Logistic Regression improved recall for churners.