https://github.com/keneandita/churnomally
Where churn meets anomaly. Churnomaly is not just a churn predictor, it's an early warning system for customer escape velocity. Blending classic churn modeling with outlier-hunting intuition, Churnomaly dives deep into behavioral signals to uncover the subtle shifts that precede goodbye. Think of it as a radar for retention.
https://github.com/keneandita/churnomally
churnomally customer-churn-prediction decision-tree-classifier knn logistic-regression ml-projects random-forest-classifier streamlit svm
Last synced: 2 months ago
JSON representation
Where churn meets anomaly. Churnomaly is not just a churn predictor, it's an early warning system for customer escape velocity. Blending classic churn modeling with outlier-hunting intuition, Churnomaly dives deep into behavioral signals to uncover the subtle shifts that precede goodbye. Think of it as a radar for retention.
- Host: GitHub
- URL: https://github.com/keneandita/churnomally
- Owner: KeneanDita
- Created: 2025-05-27T18:55:12.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2025-12-23T17:24:45.000Z (7 months ago)
- Last Synced: 2026-04-30T07:38:01.509Z (2 months ago)
- Topics: churnomally, customer-churn-prediction, decision-tree-classifier, knn, logistic-regression, ml-projects, random-forest-classifier, streamlit, svm
- Language: Jupyter Notebook
- Homepage: https://churnomally-mk9ghpwbndrjs4xzkrjb6m.streamlit.app/
- Size: 409 KB
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Customer Churn Prediction
Predicting customer churn using machine learning helps businesses identify customers likely to leave, enabling proactive retention strategies. This project includes an interactive web-based interface built with **Streamlit**, making it easy for non-technical users to explore predictions in real time.
Here is the link to the built docker image : [Docker link](https://hub.docker.com/repository/docker/keneandita/churnomally/general)
## Overview
Customer churn refers to the situation where a customer stops using a company’s product or service. In this project, I developed a machine learning solution to predict customer churn and deployed it with an intuitive **Streamlit UI**. The goal was to reduce revenue loss and improve customer satisfaction through early intervention.
Using **Python** and popular machine learning libraries such as **scikit-learn**, **pandas**, and **NumPy**, I trained and evaluated several models to find the most effective approach for churn prediction. The **Streamlit** application allows users to input customer data and receive immediate churn predictions, along with model explanations where applicable.
## Model Performance
I trained multiple machine learning models and evaluated their performance using **accuracy** as the primary metric. Below are the results:
* **Support Vector Machine (SVM):** 0.90
* **Logistic Regression:** 0.89
* **K-Nearest Neighbors (KNN):** 0.89
* **Decision Tree Classifier:** 0.885
* **Random Forest Classifier:** 0.84
The **SVM model** achieved the highest accuracy, making it the default choice in the deployed app. However, each model offers distinct advantages depending on the use case such as better interpretability or faster inference times.
## Features of the Streamlit App
* **User-friendly input forms** for customer attributes
* **Instant churn prediction output** based on trained ML models
* **Real-time visualization** of input data and prediction confidence
This interface enhances accessibility and makes it easier for business stakeholders to leverage machine learning insights without deep technical knowledge.
## Usage
1, Clone the repo(No brainer)
```bash
git clone https://github.com/KeneanDita/Churnomally
cd Churnomally
```
2, Install the dependencies
```bash
pip install -r requirements.txt
```
3, Run the server
```bash
streamlit run ./stream.py
```