Anime Recommender System

The goal of this project was to develop a collaborative-based anime recommender system capable of generating a personalized list of unique and relevent anime recommendations based on database information comprising total user history and rating ID user feedback.The source of the data is from Kaggle.com (Anime dataset from Kaggle). There are two associated datasets, rating dataset and anime dataset. The rating dataset contains 7,813,737 Ratings (Rating scale: 1- 10) from 73,516 users on 12,294 anime with a density of 0.92%; the anime dataset consists of information about each anime with 7 columns (anime_id, name, genre, type, episodes, rating, and members). Using python with SUPRISE package and leveraging custom built data cleaning and model evaluation programs, I investigated different kinds of collaborative filtering (CF) algorithms including item-based KNNWithMeans, SVD, Co-clustering, and SVDpp.

You can learn more about the results and codes at my GitHub Portfolio.