a) Consider a data set where the class variable takes 2 values and 90%
of the rows take one value (say Play=Yes) while the remaining 10% take another value (say
Play=No). We use Weka to develop a Naïve Bayes classifier and obtain the confusion matrix using
this data. Now we resample the dataset so that the new data set has 50% – 50% distribution
among the class variable (50% Play=Yes and 50% Play=No). Next we build a new Naïve Bayes
classifier using the resampled data set. What changes would you expect to see in the new
confusion matrix relative to the first model and why?