Customer Churn Prediction Using Random Forest

Project Overview

This project predicts customer churn in a telecom dataset using a Random Forest classifier. The pipeline includes data cleaning, feature encoding, model training, and evaluation with key classification metrics.

Dataset

• Contains customer demographic information, service subscriptions, billing details, and churn status.
• Target variable: Churn (Yes/No).

Data Preprocessing

• Dropped customerID as it is a unique identifier.
• Converted TotalCharges column to numeric, handling missing or invalid values.
• Dropped rows with missing data after conversion.
• One-hot encoded categorical features to convert them into numerical format.

Model Training

• Split dataset into train and test sets with stratification to preserve class distribution.
• Trained a Random Forest classifier with 100 trees and a fixed random seed.

Evaluation Metrics

• Accuracy
• Precision
• Recall
• F1 Score
• Confusion Matrix (visualized)

How to Run

1. Install dependencies:

   pip install pandas scikit-learn matplotlib
   

2. Run the script:

   customer_churn_predictions.py
   

3. The script will output evaluation metrics and display the confusion matrix.