markoleptic / SpeedyRecs

Speedy Recs is an anime recommender system devolped in RStudio using UBCF, IBCF, SVD, ALS, and a hybrid model.

SpeedyRecs: Anime Recommender System

Abstract

Speedy Recs is an anime recommender system devolped in RStudio using five different algorithms, including User-Based Collaborative Filtering (UBCF), Item-Based Collaborative Filtering (IBCF), Singular Value Decomposition (SVD), Alternating Least Squares (ALS), and a hybrid model. We obtained data from Kaggle, which was scraped from MyAnimeList, preprocessed it using RStudio, and evaluated the models for performance using various metrics.

Features

• Utilizes UBCF, IBCF, SVD, ALS, and a hybrid model for anime recommendations.
• Data sourced from MyAnimeList, preprocessed in RStudio.
• Evaluation based on error metrics, precision-recall curves, and ROC curves.
• User-friendly Shiny application interface for quick recommendations.

Previewing the Data

Error Values for Different Algorithms

Shiny App User Interface

Usage

1. Clone the repository: git clone https://github.com/markoleptic/CS5593-Project.git
2. Download anime.csv and rating_complete.csv from Kaggle and place into the root directory.
3. Install required packages: install.packages(c("shiny", "recommenderlab")) (There's probably more than just that).
4. Open the RStudio project using RStudio: SpeedyRecs.Rproj
5. Run the entire Project.Rmd file to generate all the models. This might take a while.
6. In the console, type shiny::runApp() to start the ShinyApp
7. Input your ratings to receive recommendations.

Additional File Info

• R_Program_Output.pdf: This is the resulting pdf after running Project.Rmd
• Report.pdf: A report for the project containing a full explanation, analysis, and conclusions for the project.

Contributors

• Mark Cunningham ([e-mail])
• Ethan Saloom ([e-mail])
• Branson Stickney ([e-mail])

Acknowledgments

The authors acknowledge Hernan Valdivieso for scraping anime data from MyAnimeList.