📊 Customer Churn Prediction

This project predicts whether a customer will churn (leave a service) based on various attributes like tenure, internet usage, contract type, and more. It includes complete data preprocessing, EDA, encoding, SMOTE for balancing, and model training with hyperparameter tuning.

🔍 Project Overview

• Objective: Predict customer churn using machine learning models.
• Dataset: Telco Customer Churn Dataset (commonly used for churn analysis)
• Tech Stack:
  • Python (Pandas, NumPy, Scikit-learn, Seaborn, Matplotlib)
  • SMOTE (for class imbalance)
  • Models: Decision Tree, Random Forest, XGBoost
  • GridSearchCV for hyperparameter tuning
  • Pickle for model and encoder serialization

⚙️ Workflow

1. Data Cleaning

   • Dropped unnecessary columns (customerID)
   • Replaced empty strings with NaN
   • Handled missing values in TotalCharges
   • Converted appropriate data types

2. Exploratory Data Analysis (EDA)

   • Visualized distributions (histograms, boxplots)
   • Heatmap of correlations
   • Count plots for categorical features

3. Feature Engineering

   • Label encoding of categorical variables
   • Stored encoders using pickle

4. Modeling

   • Applied SMOTE for class balancing
   • Split dataset into training and testing sets
   • Trained three models:
     • Decision Tree
     • Random Forest
     • XGBoost

5. Hyperparameter Tuning

   • Used GridSearchCV to find the best parameters for each model

6. Evaluation

   • Accuracy, Confusion Matrix, and Classification Report
   • Saved final model and encoders for future inference

📁 Files Included

• Customer_Churn_model.pkl: Trained Random Forest model
• encoders.pkl: Label encoders for categorical data
• customer_churn_prediction.ipynb: Complete Jupyter Notebook with code
• README.md: Project overview and instructions

🧪 Sample Inference

input_data = {
    'gender': 'Female',
    'SeniorCitizen': 0,
    'Partner': 'Yes',
    'Dependents': 'No',
    'tenure': 1,
    'PhoneService': 'No',
    'MultipleLines': 'No phone service',
    'InternetService': 'DSL',
    'OnlineSecurity': 'No',
    'OnlineBackup': 'Yes',
    'DeviceProtection': 'No',
    'TechSupport': 'No',
    'StreamingTV': 'No',
    'StreamingMovies': 'No',
    'Contract': 'Month-to-month',
    'PaperlessBilling': 'Yes',
    'PaymentMethod': 'Electronic check',
    'MonthlyCharges': 29.85,
    'TotalCharges': 29.85
}

🚀 Getting Started

    Clone the repository
    Open your terminal or command prompt and run:

git clone https://github.com/Mueez-lab/CustomerChurn_Prediction.git
cd CustomerChurn_Prediction

Install dependencies
(Make sure you have Python installed)

pip install -r requirements.txt

Run the notebook

    Open the Jupyter Notebook:

    jupyter notebook

    Open the customer_churn_prediction.ipynb file.

Make Predictions

    Load the trained model and encoders from the .pkl files provided (Customer_Churn_model.pkl, encoders.pkl).

    Provide your own input data in the dictionary format as shown in the notebook to get predictions.