NaN data about randomforest-matlab HOT 4 OPEN

I was getting the following error:

Warning: Do you want regression? there are just 5 or less unique values 
> In regRF_train at 163 
Error using regRF_train (line 176)
NaNs in X

this implementation cannot handle NaN data. you can try imputing the values

https://code.google.com/p/randomforest-matlab/wiki/Finding_Missing_Values

regards

Ok. Thank you.

Could you please point me in the direction of how and where it is stated in 
literature that 500 trees are the most stable for random forests, and that for 
regression the minimum leaf size is 5?

500 trees are not the most stable. they are a good enough number of trees after 
which you might find the oob error rate stabilizing

if i remember correctly, ntree and the minimum leaf size=5 for regression is 
suggested by Breiman in his paper on random forests.