There are a number of collections of open example data sets for machine learning, e.g., the one of the UC Irvine. However, they are usually organized by source or topic. Does anyone know of a list or search engine for this kind of data set, which is organized by statistical properties like number of rows/columns, scale level of target variable, rare events data, etc. ?
The UCI Machine Learning Repository that you linked actually has datasets categorized by "Default Task", "Attribute Type", and "Data Type".
For this same repository, I'd use a search engine to search the text descriptions of each dataset. For example:
If you categorizied datasets systematically, I'm sure the community would be greatly appreciative if you shared the results!
Another option is to use a programming-language specific dataset package. R's "datasets" comes to mind. Using R you can get summary information for all datasets by looping over each. More info:
library(help = "datasets")