Feature-Wizards

Movie Recommendation System

Data Pre-Processing

Ordering of files to run:

1. run remove.py to remove the duplicate entries from the dataset.
2. run data_preprocess.py to get processed movies dataset.
3. run scraper.py to scrape data from tmdb website
4. run converter.py to convert json file obtained above.

Model

1) Popularity Based:

1. run popularity_model.py to get popularity based recommendations.

2) Content-based:

1. Run content_based.py
2. run movie_similarity_based.py
3. run movie_year_analysis.py
4. run movie_era_based.py

3) Collaborative-Filtering:

1. run knn.py to identify best variation of knn
2. run surprise_model_predictions.py to test various matrix Factorization-based algorithms
3. run weight_training.py to get best parameter values for the combined_model.py which combines the above algorithms.
4. surprise_model_recs.py gives recomendaions using matrix factorization-based algorithms.

Hybrid model

run hybrid_model.py - this combines all the models to get predictions.

Files to run to get Recommendations:

1. user.ipynb to get user based recommendations
2. movie.ipynb to get movie based recommendations

Files to run to get analysis:

1. plot.py
2. plot.ipynb
3. cold_start_analysis.ipynb
4. model_hyperparameter_tuning.py