This repository contains all of the materials for my second Capstone Project for Springboard's Data Science Bootcamp.
Goodreads is a social media platform that allows users (aka readers) to rate and review books as well as see what their friends are reading, rating, and reviewing. Readers can store books on ‘shelves’ based on what they’ve read or what they want to read.
Currently, the Goodreads recommendation system uses a reader’s shelves to suggest books they might be interested in reading next. However, the system does not always offer helpful suggestions since most of the books it recommends are obscure and do not appear to be based on what they've previously rated or what their friends have read.
Is it possible to create a more useful recommendation system for readers?
To answer this question I will be using the Goodbooks datasets provided by Github user zygmuntz.
The Goodbooks dataset includes over six million ratings of ten thousand books on Goodreads. It is separated into three different files
- Ratings: Contains nearly 6 million user ratings from 53,424 users
- To-Read: Contains nearly 1 million books that users added to their 'to-read' shelf
- Books: Contains all of the meta data for 10,000 books. The metadata includes: title, author, number of ratings, number of each type of rating, and more
- Simple Model
- Collaboration Model
- Singular Value Decomposition Model
- Report
- Complete code
- Presentation