There are not enough features (rows) in the input X to generate random subsets of length { ,number,integer}. RANDFEATURES should be used preferably when subset size is much smaller than the number of features in X


randfeatures