🏦 Bank Customer Churn Prediction :-

🧠 Project Description:-

This project aims to predict whether a customer will churn (leave the bank) based on demographic and account-related features. It uses machine learning models trained on real-world customer data to help banks take proactive measures to retain valuable clients.

🌐 Live Demo

Try the app now on Hugging Face Spaces :-

👉 https://huggingface.co/spaces/YamenRM/Churn-Bank-Model

📌 Project Overview :-

Customer churn is a major issue in the banking industry. This project uses machine learning to help predict which customers are likely to churn based on historical data.

The model was trained on features like:

• Geography
• Gender
• Credit Score
• Age
• Tenure
• Balance
• Number of Products
• Has Credit Card
• Is Active Member
• Estimated Salary
• Exited (Target variable: 1 = churned, 0 = stayed)

The app allows users to input customer data and receive a prediction instantly.

📊 Dataset :-

• Dataset used: https://www.kaggle.com/datasets/kartiksaini18/churn-bank-customer

🛠️ Built with :-

• 🧠 Scikit-learn (Random Forest Classifier)
• 📊 Pandas, NumPy, Seaborn, Matplotlib
• 🌐 Streamlit for interactive web app

💻 App Features :-

Interactive sliders and dropdowns for user input

Instant churn prediction

Visual feedback on input values

Clear mapping between model features and UI

⚙️ Preprocessing :-

• Dropped unnecessary columns: RowNumber, CustomerId, Surname
• Handled categorical variables:
  • Used OneHotEncoder on Geography
  • Used LabelEncoder on Gender
• Scaled numerical features with StandardScaler

🧪 Model Used :-

• Logistic Regression
• Desicion tree
• Random Forest (best performing model)
• K-Nearest Neighbors

✅ Best Reasult :- Random Forest with GridSearchCV

• Accuracy: 0.85
• Precision (churn): 0.63
• Recall (churn): 0.61
• F1-score (churn): 0.62
• Non-churn (f1 , precision , recall) : 0.91

📈 Feature Importance :-

Identified important features influencing churn, including:

• Age
• Balance
• Is Active Member
• Geography (Germany)
• Credit Score

🚀How to Run Offline :-

• Clone the repo

git clone https://github.com/YamenRM/Churn-Bank-Model.git

cd Churn-Bank-Model

• Install dependencies

  pip install -r requirements.txt
  

• Run the main notebook

  • Model.ipynb

Or

You can launch the app locally :-

streamlit run app.py

👨‍💻 Author :-

• YamenRM , 3RD Year AI ENGINNER at UP , Gaza

stay strong 💪💪