University of Illinois Urbana-Champaign

Extracting Latent Variables Related to User Preferences on Genres in Yahoo! Music Ratings Data

Chang, Xiaowen; Bokhari, Ehsan

Content Files
Chang_Xiaowen.pdf

Permalink

https://hdl.handle.net/2142/77703

Description

Title
Extracting Latent Variables Related to User Preferences on Genres in Yahoo! Music Ratings Data
Author(s)
  • Chang, Xiaowen
  • Bokhari, Ehsan
Contributor(s)
Bokhari, Ehsan
Issue Date
2015-05
Keyword(s)
  • Statistics
  • Multidimensional Scaling
  • Principal Component Analysis
  • Nonnegative Matrix Factorization
  • Yahoo! Music
  • Latent Variables
Date of Ingest
2015-05-14T18:14:53Z
Abstract
The objective of the study is to learn community-based user preferences and identify user tastes in music via Yahoo! Music User ratings. The dataset consists of over seven million ratings of 136 thousand songs given by 1.8 million users. And it is characterized by artist, album and genre attributes. In our study, we only used a subset of this raw dataset which contains the average ratings for 4,640 users across seven popular and representative genres. We conducted multidimensional scaling, principal component analysis and nonnegative matrix factorization to extract latent variables related to user preferences on different genres.
Type of Resource
image
Genre of Resource
Conference Poster
Language
en
Permalink
http://hdl.handle.net/2142/77703
Copyright and License Information
  • Copyright 2015 Xiaowen Chang
  • Copyright 2015 Ehsan Bokhari

Owning Collections

Undergraduate Research Symposium 2015 PRIMARY