1,996 followers
典型的な医薬品を想定してるので、違う方向の製品だと別の方法を探さなければならないかも サンプルサイズ推定だと、目標が仮説検定の検出力じゃなくて予測モデルの精度だったりする その辺は探せば一応論文もあったけど…… https://t.co/g94v0JVr6D
So far, Altmetric has seen
611 X posts from 389 X users, with an upper bound of 1,519,403 followers.
13,013 followers
@rlmcelreath Here's a paper about R² & overfitting / sample size https://t.co/eqtqLcWs4f Also, as based directly on the likelihood function, R² helps quantify added value of a predictor https://t.co/Ouw3eHle1X - again, with other measures like net be…
3,787 followers
@DrSpratticus @jryckman3 Here is another article on sample size necessary for prediction models h/t @Richard_D_Riley @f2harrell TRIPOD-AI comes to mind also. https://t.co/01ijDEG8FZ
30,429 followers
@AleksanderMolak @stephensenn To well-estimate that quantity for a variety of X requires the usual sample size for developing predictive models. For the binary case: https://t.co/vGiQSGbNwv Extremely rough estimate: 96 + 15p for binary Y, where p=numbe…