top | item 45512260 (no title) svg7 | 4 months ago You just evaluate it against whatever test data you used and compute a bunch of metrics. You decide to use the model, if "bad things" happen at an acceptable enough rate. discuss order hn newest No comments yet.
No comments yet.