🧠 Stroke Prediction Pipeline 🚀 | MLOps at Scale

Welcome To The Stroke Prediction Pipeline Project! This Project Covers All Stages Of The ML Lifecycle—From Data Ingestion To Model Serving In a Production-Ready Pipeline. With CI/CD Automation using Github Actions, MLFlow Tracking, and DVC Pipeline Orchestration, This Repository Demonstrates an End-To-End MLOps Framework Designed To Tackle Real-World Machine Learning Challenges Efficiently

🔍 Overview

This Project Builds a Logistic Regression Model To Predict The Likelihood Of Stroke Based On A Dataset. With a Modular Pipeline Architecture and MLFlow Tracking, It Captures The Complete Process Of Building, Training, Evaluating, and Deploying The Model. Data is Pulled Directly from AWS S3, and The Model Is Served On a Flask Server Hosted on AWS EC2.

🌐 Live Model Endpoint:

A Website Developed With HTML And CSS, Hosted On A Flask Server, Ensures That All Inputs Fall Within The Recommended Range. Stroke Prediction Site

---

Implementation of Google Pipeline Diagram for Project Workflow

🚀 Key Features

🛠️ End-to-End Machine Learning Pipeline

• DVC (Data Version Control) to manage the ML pipeline
• AWS S3 used for storage of datasets and intermediate artifacts
• MLFlow to track experiments, metrics, and parameters

✅ CI/CD with GitHub Actions

• Automated testing using Pytest and Tox
• Deployment pipeline ensures continuous integration and delivery with every code change

📊 Comprehensive Evaluation Metrics

Model: Logistic Regression

Metric	Precision	Recall	F1-Score	Support
Class 0	0.83	0.78	0.80	2939
Class 1	0.79	0.84	0.81	2895
Accuracy			0.81	5834
Macro Avg	0.81	0.81	0.81	5834
Weighted Avg	0.81	0.81	0.81	5834

F1 Score: 0.8105

---

🧪 Model Development Stages

Exploratory Data Analysis (EDA)

• Extensive EDA to understand data distributions, patterns, and outliers
• Feature selection to retain only the most relevant predictors

Preprocessing & Pipeline

• Handled missing values and data imbalances
• Created pipelines for data ingestion, preprocessing, feature extraction, and model training

Model Training & Evaluation

• Model: Logistic Regression
• Tracked hyperparameters and metrics using MLFlow
• Stored datasets and models with DVC for seamless version control

Deployment on AWS EC2

• Hosted the Flask app on AWS EC2 with the model endpoint ready for predictions

---

⚙️ CI/CD Pipeline

🚀 This project uses GitHub Actions to automate testing, linting, and deployment.
Every push triggers the following:

• Code Quality Checks: Run Tox for testing across environments
• Automated Testing: Use Pytest to verify functionality
• Deployment to EC2: If all tests pass, the app is redeployed to the AWS server

---

🧑‍🔬 Testing

The project uses Pytest for unit tests and Tox for cross-environment testing.

🎯 Key Technologies Used

• Machine Learning: Logistic Regression
• MLOps: DVC, MLFlow
• Cloud: AWS EC2, S3
• CI/CD: GitHub Actions
• Testing: Pytest, Tox
• Website: HTML, CSS
• Server: Flask + Gunicorn

🛡️ Results

The model achieved 81% accuracy, with balanced precision and recall for both classes, making it a reliable predictor for stroke detection.