SubPop is a command line tool for finding differences between tabular datasets. It identifies combinations of values, or itemsets, that occur in one population but rarely or never occur in another.
For example, given a dataset of edible and poisonous mushrooms, SubPop can identify combinations of characteristics that distinguish the two groups. It reports each pattern along with its support: the fraction of records in that population containing the pattern.
By default, values retain their column identity, so a value in one field is distinct from the same value in another. SubPop can also treat each record as a bag of values when column identity isn’t relevant.
SubPop reads delimited text from files or standard input and lets you select the classes to compare and the minimum support for reported patterns. It is available through Homebrew, Docker, and GitHub releases.
