Problem with string type categorical data as predictor in classregtree

I understand from previous answer that predictor value in classregtree must be numeric matrix. BUT I am doing works on discretization whereby the data in the predictor value is in form of range e.g (-inf-5.20] (5.21-6.30] ... (8.91-inf) In other words, it is in strings. Q:How am I going to go about this problem since I need to use function classregtree and test to determine the accuracy of the discretization.

Thanks