Two small Python/pandas learning projects that explore correlation-based recommendations using public book and movie metadata.
src/books_recommender.pyranks related book titles and authors from a Goodreads-style dataset using average ratings.src/movies_recommender.pyranks related movie titles and directors from an IMDb-style dataset using movie ratings.
These are educational demonstrations rather than production recommendation engines. They do not model individual user-rating histories or implement collaborative filtering.
Python 3.10+ and pandas.
python -m pip install -r requirements.txtpython src/books_recommender.py books.csv
python src/movies_recommender.py IMDB-Movie-Data.csvEach program displays the available indexed items and asks which title and creator should be used as the recommendation seed.
The repository includes the historical CSV datasets used by the original learning project. Dataset rights remain with their respective data providers; the MIT license in this repository applies to the original source code and documentation, not third-party datasets.
The demonstrations intentionally use a small subset of each dataset and simple Pearson correlations over pivot tables. Sparse metadata and aggregate ratings can make the results unstable or less meaningful than modern recommender approaches.
Source code and original documentation are released under the MIT License.
Copyright (c) 2026 Mohamed Dawood.