Ricardo Cristian Ramirez
2013-Sep-18 11:32 UTC
[R] Multi multivariate feature selection problem
I have a data set with 20K variables. I want to apply Boruta feature selection but I get cannot allocate vector of size error. Hence, I decided to use bigmemory like below script but it seems that Boruta does not accept a pointer. How can I apply feature selection to 20K variables? df <- data with 20K variables df2 <- as.big.matrix(df)Bor.d <- Boruta(xyz~., data=df2) BTW, I have 4G memory and ulimit output is unlimited. I also have tried to use FSelector with same data, it works for 2K variable but it gives Seg Fault for 4K variable data. [[alternative HTML version deleted]]