When we do text classification using ML methods such as SMO in WEKA for unbalanced classes, e.g., if we have a table with a 95% value of 0 in the output column, then SMO always chooses 0 and "succeeds"with 95%.

What can we do in order to do real learning with better results.

More Yaakov HaCohen-Kerner's questions See All
Similar questions and discussions