I am using random forest in R, and for different test sets I'm using patterned seeds in the sample function. For one test set I use 1234, for the next 2345, the next 3456. I suspect it is unlikely, but could this introduce bias of some sort?
Asked
Active
Viewed 23 times
0
-
1I would be more concerned about trying so few seeds than by having some simple rule to generate them. – Galen Mar 06 '24 at 23:09
-
2The computer won't even recognize such decimal patterns because the seeds are represented in binary. – whuber Mar 06 '24 at 23:32
-
@Galen I should have put an ellipsis, this was just to show you the series. – Pierre Raynerd Mar 06 '24 at 23:55