CODEWITHSUNDEEP
rrandom-forest
Naveenan
Naveenan

Reputation: 345

difference between sampsize and classwt random forest

Can some one explain the difference between sampsize and classwt in random forest part of R package. Is there any relation between these two parameters (assuming problem is a classification problem)

Thanks

Upvotes: 3

Views: 1855

Answers (1)

Eli Sadoff
Eli Sadoff

Reputation: 7308

sampsize represents the sample size of the that is drawn. So, sampsize = 10 would mean that 10 samples are drawn. classwt is weighting of the classes. This can be used for a weighted random forest. For example, you can use this to weight more recent data as opposed to older data (see here). Having said that, downsampling (which is reducing the sample size), is seemingly as effective as weighting the classes (see here), so you can, but needn't use both.

Upvotes: 1

Related Questions