Hello
Im using RF_MexStandalone-v0.02-precompiled for classification. I was wondering
how we can modify the variable "sampsize" so that the sampling with replacement
carry out according to prior knowledge of class distribution. For instance, if
the samples are distributed in classes with this proportion [0.3 0.7]
corresponding to classes [-1 1], I want the data in each class is sampled
according to the prior knowledge. Is sampsize what i need to look into it? If
so, what would be the correct sampsize here?
Thanks
Original issue reported on code.google.com by m.saleh....@gmail.com on 24 Jul 2013 at 8:31
Original issue reported on code.google.com by
m.saleh....@gmail.com
on 24 Jul 2013 at 8:31