Book-Recommender-System - Overview

• This project has two main objectives.
• First objective is to show Top 20 books of taken dataset.
  • This is Popularity Based Recommender System
• And the second objective is to recommend 5 books to user entered book.
  • This is Collaborative Filtering Type of recommendation
  • And in this project the major concern was to focus on user ratings and make recommendation based on that
• At the beginning data merging and data cleaning was performed.
• Streamlit'framework was being used and web app was created.

Data

https://www.kaggle.com/datasets/arashnic/book-recommendation-dataset

• There is 3 csv files available for the project.
  • Books.csv
  • Ratings.csv
  • Users.csv
• After downloading 3 of them, you can start cleaning and building a model.
• In this project, Cosine Similarity was used for building collaborative filtering recommender system.

Code and Resources Used

• Python Version: 3.9
• Packages: Streamlit
• For Web Framework Requirements: pip install -r requirements.txt

Conclusion

• Both Popularity and Collaborative firterling are working at some extend.
• In collaborative filtering, cosine similarity pays huge role and it's results were more consistent.
• Here is some of the pictures from the web app.