Hacker News new | past | comments | ask | show | jobs | submit login
Show HN: Datasets.co – Share and discover new machine learning datasets (datasets.co)
81 points by mrborgen on March 20, 2016 | hide | past | favorite | 15 comments



A couple things I would suggest. Besides places like Stanford's SNAP, I would check out Socrata's OpenData portal [0], which is a place where anyone can post datasets, and the Open Data Network portal [1], which searches across all of the Socrata city portals...it's stunning what's out there. The NYPD's release of felony data is nice (if many years late), but it's nothing compared to what cities such as Chicago, Dallas, and Los Angeles have, in terms of quantity of records and detail (and the NYPD's stop-and-frisk data -- which the ACLU forced them to release -- is also much more voluminous).

I would consider providing numbers about the datasets in the default view, such as number of observations and variables. That's probably the biggest weakness, IMO, of current data portals (including data.gov)...you have to click through every link to then find out there's not much data in the set. In your situation, this applies to several of the things you've included...that Gun Ownership and Crime Rates set, for example...unless I'm missing something, but that has fewer than 40 observations, and relies on highly questionable numbers from the FBI [2] on a nationwide level...that can't possibly be of any use in a machine learning context, can it? I'm surprised it's even the basis for an academic paper (though kudos to the authors for posting their work). If you still think it's worth keeping that dataset, it'd be nice to know # of observations before having to click through.

[0] https://opendata.socrata.com

[1] http://www.opendatanetwork.com/

[2] http://www.jsonline.com/watchdog/watchdogreports/fbi-crimere...


Was a bit quick comment last time. Yes, having the amount of observations is a very natural feature to add. Shiuld be fairly simple as well, so it'll be added soon!


Thanks for the input! The variables are already described, if it was the features you were thinking about?


I was skeptical when I first saw this because there seem to be many other efforts to collect datasets.

But most of those efforts are pretty poor because they're unnecessarily difficult to understand.

I like how the data is clearly described with "Feature/Description/Example", that gives me most of what I want to know at a first glance.


Thanks! Would you also be interested in having a 'data preview' feature, where you'd see a tiny part of the data in the browser?


It would be nice to have some estimate of the data size (e.g. number of rows).


Hey, on datasets added from now on, this'll be an option. Check it out in action here (displayed under the image).

http://www.datasets.co/dataset/All-UFC-Fights

Was that what you were thinking?


Yes, it's what I meant.

But in this case there are 19 features (as said) or 10 features (no of columns)?


It's avtually 19, I just didn't understand all of them to be able to write proper descriptions.


Then it is better to write all columns, even if some with "(unknown)" desc.


Hey all!I built this site because I think it should be easier and more fun to discover new datasets. I'd love hear your thoughts, suggestions and critisism!


Awhile ago I put together a list of big (100,000+ rows) and public datasets...not all of them are ideal for machine learning applications but you'll probably find a few worth sharing:

http://cjlab.stanford.edu/2015/09/30/lab-launch-and-data-set...


Awesome list, thanks!


Good datasets source .. one point - you can register an empty data set . You might want to fix that.. rest all looks great .. good job


This is awesome, as someone who is currently playing with ML this is a great resource for playing with Datasets. Thanks to everyone involved in this project. Keep it up.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: