//
sign in
Post
by @danabra.mov
PostEmbed
by @danabra.mov
Record
by @jimpick.com
Record
by @atsui.org
+ new component
Post
I would say that Step 2: Get to know your data, should be Step 0, though; even if you end up not doing ML, this step is necessary. And I'd have liked to see this ceveat for feature selection: --> Never use the parameter you aim to predict for feature selection with *the whole dataset*! <-- (1/3)