Stratified Sampling about randomforest-matlab HOT 6 OPEN

oh i am so sorry. i sortof missed this issue. 

let me look at it and get back to you in a couple of days.

hi, is the internal oob error process is stratified? 

i have the following six classes with the available number of samples. ( with a 
372 feature vector)

class1 150 
class2 102 
class3 109
class4 94 
Class5 208 
Class6 58  total samples 721.

for testing i have used 5 fold stratified cross validation. 

Any special suggestions regarding my data? 
Is I have something to do with the classwt or cutoff option?

regards,
Ihtesham

there is one more related questions . in your demo code you write 
extra_options.strata, is this related to my question in comment 2.

regards,
Ihtesham

hi @Ihtesham and the OP

sorry for my late reply.

i am still not finished debugging the strata code, i should have something by 
the end of this week. thanks

@Ihtesham, yup it has to do with the extr_options.strata (not yet done) where 
sampling can be done according to how the labels are distributed.

thanks for your valuable reply. I am waiting for an update. 

hello,

i added support for stratified sampling in the svn source. Look at the 
tutorial_ClassRF.m file for the last two examples for how to do stratified 
sampling.

i hope it works. do tell me if you encounter any problems or have any 
questions. 

thanks