spearmanr

I'd like to compute spearmanr in datatable for a given data frame, giving its self spearmanr (as opposed to one dataframe on another). I think it might already be possible. Here's codified way it could be done, using some other package, but that could be inherited by datatable:

https://deepgraph.readthedocs.io/en/latest/tutorials/pairwise_correlations.html

And some notes about parallel performance and avoiding the double argsort at:

https://stackoverflow.com/questions/52371329/fast-spearman-correlation-between-two-pandas-dataframes

The actual calculation is very simple, but wonder if you have thoughts.

Start by reviewing the linked DeepGraph pairwise-correlations tutorial and the Stack Overflow notes on parallel performance and avoiding duplicate argsorts. The requested outcome is a self-Spearman correlation computation for a datatable data frame, with the API and performance approach established.

Vom Indexierungsmodell aus dem Issue-Text verfasst.