classification performance metrics (sensitivity specificity ) showing 0 or NaN value

Hello All, I am doing 10 fold CV method to classify binary data. While evaluating the performance of the classifier using standard formula, many times I found 0 or NaN value for different metrics (such as, TPR, FPR, MCC value, precision, recall etc). How can I avoid 0 or Nan value to get correct rate? Any idea or links? Thanks

[Greg Heath]

First find out out what is causing them.