convert numpy array to Matrix rpy2, Kmeans

问题:

I have a numpy 2D array self.sub and i want to use it in rpy2 kmeans. k = robjects.r.kmeans(self.sub,2,20) i always get the following error: valueError: nothing can be done for the type at the moment! what can i do?

回答1:

From the rpy2 docs, R matrices are just vectors with their dim attribute set. So for a numpy two-dimensional array x

import rpy2.robjects as robj

nr, nc = x.shape
xvec = robj.FloatVector(x.transpose().reshape((x.size))
xr = robj.r.matrix(xvec, nrow=nr, ncol=nc)

You have to transpose the numpy array because R fills matrices by columns.

Edit: Actually, you could just set byrow=True in the R matrix function, and then you wouldn't need to transpose.