> I have a dataset which has output variables True or False.. How can I convert it to numeric such that I get RMSE values rather than Accuracy or AUC.
RMSE (root mean squared error) is calculated for classification as well.
Converting T/F to 1/0 (or something like that) changes the discrete
classes to a continuous regression problem and interpreting
predictions (-0.1? 9.1? 0.3? 0.51?) may become a problem.
In terms of your query: You could just use a text editor to replace
the strings. Otherwise, use the RenameNominalValues filter and then
change the attribute type in the ARFF file header to "numeric" (using
a text editor).