Center for Machine Learning and Intelligent Systems
About  Citation Policy  Donate a Data Set  Contact

Repository Web            Google
View ALL Data Sets

Repeat Consumption Matrices Data Set
Download: Data Folder, Data Set Description

Abstract: The dataset contains 7 datasets of User - Item matrices, where each entry represents how many times a user consumed an item. Item is used as an umbrella term for various categories.

Data Set Characteristics:  


Number of Instances:




Attribute Characteristics:


Number of Attributes:


Date Donated


Associated Tasks:


Missing Values?


Number of Web Hits:



Dimitrios Kotzias, dkotzias '@', University of California Irvine

Data Set Information:

There are 7 datasets from Reddit, Twitter, Gowalla and Lastfm.
Each matrix contains how many times a user 'consumed' and item. Items can be locations, artists, or subreddits.
Details about each dataset are presented below. (In the parenthesis is the number of Users x Items)

tw_oc (13k x 11k): tweets with geolocation from Orange County CA area. Items are locations a user visits in this case.
tw_ny (30k x 11k): Same as tw_oc but from the New York area.

go_sf (2k x 7k): Check-ins from the app Gowalla, from the San Fransisco area. Full dataset here: [Web Link]
go_ny (1k x 7k): Same as go_sf, but from the New York area.

lastfm (992 x 15k): How many times, a user listened to each artist. Covers 3 years of listening habbits, full dataset here: [Web Link]∼ocelma/[Web Link]

reddit_top (113k x 21k): How many times a user posted in a subreddit. These are the 130k most active users from 2015 and 20k most subscribed subreddits. This dataset is very large and can take a lot of time to load/use.
reddit_sample (20k x 21k): Same as reddit_top, but a sample of 20k users.

Attribute Information:

The attributes represent items (categories) that uses tend to select multiple times. These can be music artists, subreddits or locations on the map.

Relevant Papers:

Predicting Consumption Patterns with Repeated and Novel Events by Dimitrios Kotzias, Moshe Lichman and Padhraic Smyth.

Citation Request:

If you have no special citation requests, please leave this field blank.

Supported By:

 In Collaboration With:

About  ||  Citation Policy  ||  Donation Policy  ||  Contact  ||  CML