Fastest Multivariate KDE Method

Question

I have a dataset of ~3x10^7 points in five dimensions, for which I must calculate:

• One KDE model for the total dataset;
• Fifty models for a subset of 1% of