I'm writing a paper about an algorithm. The algorithm is qualitatively evaluated using empirical tests on sample data. I find it difficult to find a good word for multiple such data. It should be possible to put the word which describes this in plural form, describing multiple sets of multiple such data.

To be more precise:

My algorithm takes some input data (a "problem instance") and generates an output. Similar problem instances are grouped together, and for that I am looking for a word to describe this. And also, I need to be able to talk about multiple such groups: they have different properties, but within one such "group" the instances have similar properties.

For example (in which I call it "sets of instances"):

We generated the following "sets of instances". Primarily, we need different sizes of inputs, so one "set" contains small instances, the second one medium-sized instances, and the third one large instances.

Note: the input data is automatically generated randomly, and as such artificially created (i.e. not an extracted sample of real data).

  • What about "body or collection of samples" ? Also "assortment".
    – Graffito
    May 5, 2016 at 19:53
  • 1
    Other possibility: "test-data series".
    – Graffito
    May 5, 2016 at 20:34

2 Answers 2


In my field we typically refer to these as "datasets," although it may still be more frequent to term them "data sets" in the broader scientific literature.

  • Thank you for your answer. Is the term "test set" also used, when the data is for testing / experimental purposes? Because to me, "data set" sounds quite generic.
    – leemes
    May 5, 2016 at 20:12
  • 1
    We tend to subsample or group datasets -- or results -- and simply describe how that was done. But, yes, if a group of similar datasets (real or randomly generated) needs a label then "test set" should work (or "test group"). Then I presume you describe the results of the qualitative assessments, summarizing within-group runs.
    – KWinker
    May 5, 2016 at 20:50

You describe the datasets used for eval as randomly generated. Aggregated is a very handy word in these circumstances since the eval isn't repeatable due to the Monte Carlo aspect of the input datasets. Your qualitative eval would be based on an aggregate of the results of a series of runs. You would usually want to regard the input datasets in the aggregate in such circumstances also. Otherwise, you probably shouldn't be using randomly generated inputs for trials.

Idioms 11. in the aggregate, taken or considered as a whole:

aggregated. (n.d.). Dictionary.com Unabridged. Retrieved May 05, 2016 from Dictionary.com website

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.