Closed Klopfe closed 2 years ago
See if one can improve the computation of X[:, current_cluster] @ sign_w[current_cluster] performance for sparse matrices. See profiling below
I think our best option might be to construct a sparse X_reduced matrix which would contain the aggregated columns of X for each cluster with corresponding signs. What do you think @mathurinm @jolars ?
See if one can improve the computation of X[:, current_cluster] @ sign_w[current_cluster] performance for sparse matrices. See profiling below