I am searching for a multivariate dataset consisting of vectors (no time series) for clustering experiments. It should have about 8-20 numeric features (non-numeric would have to be ignored by me) and about 100-500 instances. Having some labels additionally might be useful, but is not required.

1 Answer 1


I have worked with the 2011 Irish Census dataset for clustering and demographic classification previously and found it to be very useful.


There are various aggregated levels you can download the data, some of which are in your 100-500 instances range. There is also a fairly comprehensive data dictionary available.

There are no labels but mapping your results (boundary files are also available) would provide some sort of clarification (you wouldn't expect a cluster with high proportion of farmers to be in an urban area).

  • Thanks! Which algorithms did you use? I know how my algorithm works and I am pretty firm in clustering in general, so I should be able to predict whether my algorithm will work as well if you tell me. Also, do you have a specific set (a specific CSV) that you used, you know worked well for demonstration and you could upload?
    – Make42
    Jan 19, 2018 at 17:07
  • I used k-means clustering after performing principal component analysis. As for a CSV of the specific data I used, I'm afraid this has been lost to time but I will try to root it out and update if I manage to do so
    – M4GU1R3
    Jan 19, 2018 at 19:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.