how to predict new data in unsupervised forest

[mytarmail]

how to predict new data

library(randomForestSRC)
m <- matrix(rnorm(1000),ncol = 5)
o <- sidClustering(m, k = 5)
o$clustering

predict??

[ishwaran]

Hi, the sidClustering function is designed for clustering and not prediction. If you want to fit a multivariate RF for prediction purposes, then you should use the rfsrc function. See the following vignette for some examples of how to use multivariate RF

[mytarmail]

Hi, the sidClustering function is designed for clustering and not prediction. If you want to fit a multivariate RF for prediction purposes, then you should use the rfsrc function. See the following vignette for some examples of how to use multivariate RF

Hello! I understand that this is a function for clustering, but in the vast majority of clustering algorithms there is a prediction method, this is a normal practice, so I thought that it should be here. In any case, thanks for the reply.

[ishwaran]

Yes, that's true that you can do prediction with many clustering algorithms. The issue here is that sidClustering does a special processing of the X-features before it fits a random forest model. Because of this, it's difficult (but not impossible) to set it up to do prediction. I will look into this and maybe in the future put this option in. Thanks for your input.