A Nazi's wet dream. In the submitted article, they say that 10% of the data was used as test data set. I cannot find that info in the orig article. After skimm

A Nazi's wet dream. In the submitted article, they say that 10% of the data was used as test data set. I cannot find that info in the orig article. After skimming through the paper, it seems to me the reported accuracy results from the 10-fold cross-validation approach they used - which probably explains the 10% mentioned above. They should probably have calculated the accuracy with a fresh data set. It also seems they have a certain false pos rate.