A machine learning project that analyzes YouTube video comments to determine sentiment (positive, negative, or neutral). This project includes a Flask web API and a Firefox browser extension for real-time sentiment analysis of YouTube comments.
The project features complete MLOps implementation with CI/CD pipeline, Docker containerization, and AWS cloud deployment using ECR for container management, S3 for data storage, and EC2 for hosting.
This project was built following a tutorial with some personal modifications. It uses machine learning to classify YouTube comments into different sentiment categories and provides both a web API and browser extension interface. The project demonstrates end-to-end ML deployment with modern DevOps practices including automated testing, containerization, and cloud infrastructure.
- Sentiment Analysis: Classifies YouTube comments into positive, negative, or neutral sentiments
- Flask Web API: RESTful API for comment sentiment prediction
- Firefox Extension: Browser extension for real-time YouTube comment analysis
- Data Visualization: Word clouds and sentiment distribution charts
- Machine Learning Pipeline: Complete ML pipeline with preprocessing, training, and evaluation
- Backend: Flask, Python
- Machine Learning: LightGBM, scikit-learn, TF-IDF vectorization
- Data Processing: pandas, NumPy, NLTK
- Visualization: matplotlib, seaborn, WordCloud
- MLOps: MLflow, DVC
- Frontend: HTML, CSS, JavaScript (Firefox Extension)
- Deployment: CI/CD, Docker, AWS
The project includes Docker containerization for easy deployment and consistent environments across different platforms. The Dockerfile provides a complete setup for running the application in a containerized environment.
Continuous Integration and Continuous Deployment pipeline is implemented using GitHub Actions, enabling automated testing, building, and deployment of the application.
The project is configured for AWS deployment with:
- S3: For storing model artifacts and data files
- EC2: For hosting the Flask application
- ECR: For storing and managing Docker container images
The project uses a LightGBM classifier with TF-IDF vectorization for text processing. The model was trained on Reddit comment data and fine-tuned for YouTube comment analysis.
This project was developed following a tutorial with additional modifications and improvements. Special thanks to Bappy and freeCodeCamp.
Tutorial Reference: YouTube Sentiment Analysis Tutorial