Harness the wisdom of the crowd to find movies you'll love! Playpick is a persistent, multi-user web app that delivers personalized movie recommendations using collaborative filtering techniques.
- Persistent recommendations: Your preferences are saved for consistently tailored suggestions.
- Collaborative filtering: Uncover hidden gems based on the ratings of like-minded users.
- Streamlit frontend: Enjoy a user-friendly and interactive experience.
- MongoDB integration: Seamless data storage and retrieval.
- Model optimization: Quantized model for efficient deployment within Streamlit Cloud's 1GB limit.
- Performance tracking: Average Precision@20 of 0.3795 and Average Precision@100 of 0.4604.
- Explore the app live: https://pick-1-movie.streamlit.app
- Data Collection: Aggregates latest IMDb movie dataset.
- Data Cleaning: Removes obsolete movies and insignificant users for enhanced model accuracy.
- Correlation Matrix: Uncovers patterns in user ratings using a pivot table.
- Model Quantization: Reduces model size by half for efficient deployment.
- Performance Evaluation: Average Precision@k metrics ensure recommendation quality.
PlayPick Interface:
Evaluation Metrics:
Metrics: Avg Precision@20=0.3795 Avg Precision@100=0.4604
