# Customer Churn Prediction Predict customer churn using machine learning on the Telco Customer Churn dataset. This project uses a Random Forest Classifier with SMOTE to handle class imbalance and identify customers who are most likely to leave a telecom service. Alternative techniques such as class_weight= 'balanced' or cost-sensitive learning should also be considered. Project Overview This project follows a complete machine learning workflow: Cleaned and preprocessed the dataset ,Handled missing values , Encoded categorical features ,Scaled numerical features , Balanced the training data using SMOTE ,Trained a Random Forest Classifier ,Evaluated the model using multiple performance metrics & Identified the most important features influencing customer churn Technologies Used: Python ,Pandas ,NumPy ,Scikit-learn ,imbalanced-learn (SMOTE) ,Matplotlib & Seaborn Model Performance Overall Performance Metric Score Accuracy 76% ROC AUC 0.812 Churn Class Metric Score Precision 0.54 Recall 0.64 F1-Score 0.59 Non-Churn Class Metric Score Precision 0.86 Recall 0.81 F1-Score 0.83 Results Confusion Matrix  ROC Curve:  Feature Importance:  Key Insights The model found that the most influential features for predicting customer churn were: Customer tenure ,Total charges , Monthly charges ,Contract type & Payment method (Electronic Check) Customers with shorter tenures, higher monthly charges, month-to-month contracts, and electronic check payments were more likely to churn. Project Structure Customer-Churn-Prediction/ customer_churn.py requirements.txt confusion_matrix.png roc_curve.png feature_importance.png README.md data/ Telco-Customer-Churn.csv Installation: Clone the repository. git clone https://github.com/Acacia21-code/customer-churn-prediction.git Move into the project directory. cd customer-churn-prediction Install the required packages. pip install -r requirements.txt Run the project. python customer_churn.py Dataset: Telco Customer Churn Dataset https://www.kaggle.com/datasets/blastchar/telco-customer-churn Future Improvements: Tune model hyperparameters using GridSearchCV , Compare performance with XGBoost and Logistic Regression , Build an interactive Streamlit web application ,Add model explainability using SHAP & Experiment with additional feature engineering Author: Mbali Aspiring Machine Learning Engineer passionate about building practical machine learning solutions and continuously learning new technologies.