📉 Customer Churn Prediction System

🖥️ GUI Application: Desktop app using Tkinter
🧠 Model Type: Random Forest Classifier
💼 Purpose: Predict whether a customer will churn based on service and usage details.

---

📋 Project Overview

This project is a Python-based GUI application for predicting customer churn. It features a graphical interface for loading datasets, training a machine learning model, and making predictions on new data. The application is tailored to provide business insights by allowing users to assess which customers are at risk of leaving a service.

Key functionalities include:

• Data Loading and Exploration: Load customer datasets (CSV), inspect structure, check nulls, types, and value distributions.
• Visualization: See churn distribution and feature importance graphs.
• Preprocessing: Encodes categorical variables and scales numerical values.
• Model Training: Uses RandomForestClassifier from scikit-learn to train a churn prediction model.
• Model Persistence: Save and reload models using joblib.
• Prediction Interface: Input customer attributes (sample/random/manual) to get churn predictions with confidence levels.

---

🛠️ Technologies Used

• Python 3
• Tkinter – GUI framework
• Pandas / NumPy – Data handling
• Seaborn / Matplotlib – Data visualization
• Scikit-learn – ML model, preprocessing, evaluation
• Joblib – Model saving/loading

---

🚀 How to Run

1. Install required libraries:

pip install -r requirements.txt

2. Run the application:

python churn_gui.py

3. Load a dataset (CSV), train a model, make predictions, and visualize results.

---

📸 Screenshots

📊 Data Exploration

---

🧪 Model Training

---

🔮 Make Predictions

---

🎥 Preview

---

🎯 Features Breakdown

📁 Load CSV File

• Select and load a CSV file with customer data.
• Displays file name, shape, column data types, nulls, and unique values.

📊 Data Exploration

• Shows churn class balance (e.g., Yes/No counts and percentages).
• Displays this visually using bar charts.

⚙️ Data Preprocessing

• Converts TotalCharges to numeric, handles missing values.
• Label encodes categorical features and target.
• Separates features and labels.

🧪 Model Training

• Trains RandomForestClassifier on 80% of data, stratified split.
• Shows accuracy, confusion matrix, and classification report.
• Plots and ranks all features by importance.

💾 Model Save/Load

• Save your trained model and encoders.
• Load previously saved model to make new predictions.

🧮 Predictions

• Input new customer info manually or use sample/random fill.
• Predicts churn (Yes/No) and shows model confidence.
• Colors prediction text (green for No, red for Yes).

---

📁 File Structure

📦 AI Lab Final Project
 ┣ 📂 data
 ┃ ┗ 📄 churn.csv
 ┣ 📂 assets
 ┃ ┗ 📄 data.png
 ┃ ┗ 📄 model.png
 ┃ ┗ 📄 prediction.png
 ┃ ┗ 📄 preview.gif
 ┣ 📂 model
 ┣ 📄 churn_gui.py
 ┣ 📄 LICENSE
 ┣ 📄 requirements.txt
 ┗ 📄 README.md

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

✍️ Author

Muhammad Huzaifa Karim
GitHub Profile

---

📬 Contact

Feel free to reach out if you have any questions or feedback!
Email: karimhuzaifa590@gmail.com

---

© 2025 Muhammad Huzaifa Karim