Workflows are now clustered through KMeans where number of features is number of errors time number of sites. This seems to be too many features to me. I think separating and have number of errors + number of sites will be enough information. This is working nicely otherwise, for now.
Workflows are now clustered through KMeans where number of features is number of errors time number of sites. This seems to be too many features to me. I think separating and have number of errors + number of sites will be enough information. This is working nicely otherwise, for now.