I'm using fDMA model in R language. The input of this model is x, y. In my case,x contains 194 features. The problem is that the model constructs a matrix containing all the combinations of models that we can build with these 194 features(here, the total number of combinations is 2^194), and them make the estimations. As there is a limit in R for the size of a matrix(2^51), I get error after running the model. Is there any way to handle this problem without reducing the size of the combination matrix?
I tried with bigmemory. However as the combination matrix is constructing in the model, I was able to make the input variable(x) as a big matrix and it didn't work.(I'm not sure if I've done it correctly)