Hi @ajaybabu27, I haven't seen this error message before. But after digging around for related bug reports on statsmodels, it likely has to do something with ill-defined input.
https://groups.google.com/forum/#!topic/pystatsmodels/cWdOIQmcs4w
There are really only two input types that could be causing issues here, the input metadata or the biom table. Could you do the following?
- Can you try rerunning the model on few inputs, and report which ones crash?
- Have you confirmed that your biom table doesn't have serious problems with zeros? See here for discussion on this: Detected zero variance balances while performing "gneiss ols-regression" - #10
- If either of the two solutions aren't working, could you post all of your qzas + metadata to reproduce this error?