On 27/07/12 5:00 PM, Sani Zimit wrote:
In the quest to obtain a more reliable results, I run 10-fold
cross-validation ten times by manually diving the data set into 10-
folds using different seed for each set of 10 folds.
After running the 10-fold CV 10 times I got 10 different confusion
matrices. The average of their entries is 160.9, 40.1, 38.8 46.2.
is it appropriate if I ROUND the entries to whole numbers that is
It's probably reasonable to do this.