Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/kfrural/customer-churn-prediction

Customer churn prediction using machine learning. The project follows CRISP-DM and KDD methodologies, including data preprocessing, feature engineering, modeling, and evaluation. It also features an interactive dashboard for visualizing results.
https://github.com/kfrural/customer-churn-prediction

crisp-dm data jupyter kdd python

Last synced: 14 days ago
JSON representation

Host: GitHub
URL: https://github.com/kfrural/customer-churn-prediction
Owner: kfrural
License: mit
Created: 2024-10-23T22:03:23.000Z (22 days ago)
Default Branch: main
Last Pushed: 2024-10-24T11:35:57.000Z (21 days ago)
Last Synced: 2024-10-25T13:35:53.582Z (20 days ago)
Topics: crisp-dm, data, jupyter, kdd, python
Language: Jupyter Notebook
Homepage:
Size: 59.6 KB
Stars: 0
Watchers: 1
Forks: 0
Open Issues: 0
Metadata Files:
- Readme: README.md
- License: LICENSE

Awesome Lists containing this project

README

# Customer Churn Prediction

This project aims to predict customer churn for a telecom company using machine learning techniques. By leveraging two methodologies, **CRISP-DM** and **KDD**, we ensure a structured and insightful approach to the data mining process. The project includes data preprocessing, feature engineering, modeling, and evaluation phases, and is supported by a deployable dashboard for presenting the results.

## Project Overview

- **Objective**: Predict whether a customer will churn (leave the company) based on historical customer data.
- **Dataset**: Telecom customer churn data containing demographic information, customer behavior, and churn status.
- **Methodologies**:
- **CRISP-DM**: Cross Industry Standard Process for Data Mining.
- **KDD**: Knowledge Discovery in Databases.

## Directory Structure

```plaintext
/customer-churn-prediction
│
├── /data
│ ├── telecom_churn_data.csv # Dataset for customer churn
│
├── /notebooks
│ ├── 01_data_understanding.ipynb # Data exploration and understanding
│ ├── 02_data_preprocessing.ipynb # Data cleaning and preprocessing
│ ├── 03_feature_engineering.ipynb # Feature selection and engineering
│ ├── 04_modeling.ipynb # Model training and tuning
│ ├── 05_evaluation.ipynb # Model evaluation and validation
│
├── /src
│ ├── data_preprocessing.py # Data preprocessing script
│ ├── feature_engineering.py # Feature engineering script
│ ├── modeling.py # Modeling script
│ ├── evaluation.py # Model evaluation script
│
├── /dashboard
│ ├── app.py # Dashboard for result visualization
│
├── /docs
│ ├── CRISP_DM_overview.md # Overview of CRISP-DM methodology
│ ├── KDD_overview.md # Overview of KDD methodology
│ ├── Project_Plan.md # Project plan and methodology documentation
│
├── README.md # Project overview and instructions
├── requirements.txt # List of Python dependencies
└── LICENSE # License for the project
```

## Workflow

### 1. **Data Understanding**
The initial phase involves exploring the data to understand key attributes and relationships. The dataset includes features such as:

- **Customer Demographics**: Gender, SeniorCitizen, tenure.
- **Service Information**: MonthlyCharges, total service usage.
- **Churn Status**: Whether the customer left the company or not.

Exploration results are documented in `01_data_understanding.ipynb`.

### 2. **Data Preprocessing**
Cleaning and preparing the data for modeling. This includes:

- Handling missing values.
- Encoding categorical variables.
- Normalizing or scaling numeric data.

Details in `02_data_preprocessing.ipynb` and `data_preprocessing.py`.

### 3. **Feature Engineering**
New features are derived to improve model performance:

- Interaction features (e.g., tenure * monthly charges).
- Aggregations or groupings based on customer type.

More details in `03_feature_engineering.ipynb` and `feature_engineering.py`.

### 4. **Modeling**
Several machine learning models are tested, including:

- Logistic Regression
- Random Forest
- XGBoost

Model training, cross-validation, and hyperparameter tuning are performed to identify the best model.

Check `04_modeling.ipynb` and `modeling.py` for implementation.

### 5. **Evaluation**
The best-performing models are evaluated using:

- Accuracy, Precision, Recall, and F1-Score.
- Confusion Matrix.
- ROC and AUC metrics.

Results are recorded in `05_evaluation.ipynb` and `evaluation.py`.

### 6. **Dashboard**
A simple interactive dashboard is developed using `Plotly Dash` to visualize the model's performance, important features, and predictions.

Check the `dashboard/app.py` for the dashboard setup.

## Technologies Used

- **Python**: Core programming language.
- **Pandas**: Data manipulation and analysis.
- **Scikit-learn**: Machine learning models and utilities.
- **XGBoost**: Advanced gradient boosting models.
- **Plotly Dash**: Dashboard development.
- **Jupyter Notebooks**: For interactive data exploration and model building.

## Setup Instructions

1. Clone the repository:

```bash
git clone https://github.com/your-username/customer-churn-prediction.git
cd customer-churn-prediction
```

2. Create a virtual environment and install dependencies:

```bash
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
pip install -r requirements.txt
```

3. Download the dataset and place it in the `/data` folder (or use the example `telecom_churn_data.csv` provided).

4. Run the notebooks to explore the data and train the models:

```bash
jupyter notebook
```

5. Run the dashboard for visualizing the results:

```bash
python dashboard/app.py
```

## Results

- **Best Model**: XGBoost
- **Key Metrics**:
- Accuracy: 85%
- Precision: 82%
- Recall: 78%
- F1-Score: 80%
- **Feature Importance**: Tenure, Monthly Charges, Contract Type were found to be the most predictive features.

## Future Work

- Improve model interpretability using SHAP or LIME.
- Extend dashboard for real-time monitoring of churn predictions.
- Experiment with deep learning models for further improvements.