Customer Churn Prediction using Machine Learning & Streamlit

Overview

-This project is an end-to-end Machine Learning application that predicts whether a customer is likely to churn (leave a service) based on their demographic, service usage, and billing information.

-The model is deployed as an interactive web application using Streamlit Cloud, making it easily accessible for real-time predictions.

Live Demo

Deployed App: (https://customer-churn-prediction-xf9uobvp49mv4njxsdojxw.streamlit.app/)

Problem Statement

Customer churn is a major challenge for subscription-based businesses. The goal of this project is to:

-Predict customer churn in advance

-Help businesses take proactive retention actions

Tech Stack

Programming Language: Python

Machine Learning: Scikit-learn

Model Used: Logistic Regression

Data Processing: Pandas, NumPy

Model Deployment: Streamlit Cloud

Version Control: Git & GitHub

Containerization: Docker

📂 Project Structure

Customer_Churn_Prediction/
│
├── lg_app.py                     # Streamlit application
├── requirements.txt              # Python dependencies
├── Dockerfile                    # Docker configuration
├── .dockerignore                 # Docker ignore rules
├── Customer Churn Prediction using Logistic Regression.ipynb
│                                 # Main Jupyter notebook
├── models/
│   └── logistic_Churn.pkl        # Trained ML model + scaler
└── README.md                     # Project documentation

Dataset Description

Dataset Link: https://www.kaggle.com/datasets/rashadrmammadov/customer-churn-dataset

The dataset contains customer-level information such as:

-Gender, Senior Citizen status, Partner & Dependents, Tenure, Phone & Internet Services, Contract type, Monthly & Total Charges, Churn (Target Variable)

Machine Learning Workflow

-Data Cleaning

-Removed irrelevant columns

-Handled missing values

-Feature Encoding

-Categorical variables encoded using Label Encoding

-Feature Scaling

-StandardScaler applied to numerical features

-Model Training

-Logistic Regression trained on 11 features

-Model Serialization

-Model and scaler saved using pickle

-Deployment

-Integrated with Streamlit for real-time inference

Model Performance

-Evaluation Metric: Accuracy

Model chosen for:

-Interpretability

-Fast inference

-Production suitability

Application Features

-Interactive user input form

-Real-time churn prediction

Displays:

-Churn or Not Churn

-Actionable retention tips

-Cloud-hosted and accessible via browser

Docker Support

The application can also be run using Docker:

docker build -t churn-app . docker run -p 8501:8501 churn-app

▶️ How to Run Locally

1️⃣ Clone the repository git clone https://github.com//Customer-Churn-Prediction.git cd Customer_Churn_Prediction

2️⃣ Install dependencies pip install -r requirements.txt

3️⃣ Run the app streamlit run lg_app.py

Key Learnings

-Importance of consistent preprocessing between training and inference

-Handling feature mismatch errors in deployment

-Using robust file paths for cloud environments

-End-to-end ML lifecycle: training → serialization → deployment

👤 Author

Arpit Kumar Aspiring Data Scientist | Machine Learning | Deep Learning