Public SDV Datasets
The SDGym library includes a variety of public, demo datasets that you can use from benchmarking. These come from our main SDV library.
Available Datasets
View all the datasets that are available through the get_available_datasets
function.
get_available_datasets
See all the publicly available demo datasets that are available to use.
Parameters
modality
: A string describing the type of data. At this time, the only supported modality is'single_table'
.
Returns A pandas DataFrame object that describes the dataset name, dataset size and number of tables.
The reported dataset size is based on fully loading the data into Python. You may find slight deviations between the CSV file size and reported size.
Recommended Datasets
By default, the benchmarking includes 9 of the available datasets. These datasets were chosen as examples of rich data that you may find in real world settings. They of substantial size, contain a variety of columns and meet the SDGym standards for single table data.
Dataset | Description |
---|---|
| Attributes corresponding to real adults in the 1994 US census ↗ |
| Simulated data for an alarm messaging system when monitoring patients ↗ |
| US census data extracted from 1994 and 1995 ↗ |
| Health properties corresponding to different patients |
| Information about forest covers in different regions of the world ↗ |
| Web logs of corresponding to a random selection of Expedia users browsing the website ↗ |
| Simulated data about various student drivers and their vehicles ↗ |
| Network traffic that contains simulated attacks on a U.S. air force LAN ↗ |
| Attributes about published news articles |
Benchmarking the datasets
You can benchmark any of the publicly available datasets by providing their string names into the sdv_datasets
parameter.
Want to include your own datasets? See the Custom Datasets section for more information.
Last updated