1

Is anyone aware of a source for sample data sets with known outliers? I've been looking around for years but haven't come up with a solution, short of creating my own limited database.

Sets with known outliers according IQR, Q-test and Z-test would help me a lot.

1

My advice would be to look at the documentation for the relevant statistical procedures in popular statistics software, specifically in R or SAS. Statistical function documentation usually has sample datasets and examples of the usages. http://www.rdocumentation.org/ http://support.sas.com/documentation/onlinedoc/stat/

0

You can take a look at Wine Data Quality from the UCI Machine Learning Repository

https://archive.ics.uci.edu/ml/datasets/Wine+Quality

These datasets can be viewed as classification or regression tasks. The classes are ordered and not balanced (e.g. there are munch more normal wines than excellent or poor ones). Outlier detection algorithms could be used to detect the few excellent or poor wines. Also, we are not sure if all input variables are relevant. So it could be interesting to test feature selection methods.

0
  1. Don't use Dixon's Q.
  2. I'm having trouble understanding why you don't just make up a data set to specifications, if the goal is validation of outlier detection? A simple mixture distribution will suffice, but a data set can be constructed to have the same data issues that the expected empirical data will have.
  • Maybe point 1 would be less off-topic if you explained why? – Nicolas Raoul Dec 4 '14 at 6:04
  • Dixon's Q assumes that there is <em>exactly</em> one outlier in the data (otherwise the test distribution is not well-defined). Speaking as a statistician, I'd say that this is a very strong prior belief on the distribution. – abaumann Dec 4 '14 at 19:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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