@PhDemetri Curious on your take on this re: logistic reg vs more complicated models, but this paper compares a few models including logistic reg to detect blood tube errors https://t.co/swuqKBd5ub
@jrosell In vscode click “split editor right”, then right click on the right side editor header and select “split down”, then open an r terminal, drag the terminal name “R interactive” up into the new space, then ctrl+j to hide bottom terminal
@PhDemetri I’m working on a clinical model in a hospital lab setting and am thinking thru something similar. I’m leaning towards a C5.0 rules model, it’s almost as performant as xgboost but you can get plain English rules for each prediction in addition to var imp