I'm writing a proposal for a prediction model predicting BP (continuous outcome, predicting trend over time). For assessing model performance, I'm seeing discrimination and calibration as the most commonly reported measures. Calibration makes sense, but I'm not sure if discrimination is applicable for a continuous outcome - you can't separate those who do or don't have an event, so you can't assess how well the model discriminates between them. Am I correct in thinking this? If so, are there any other measures that I can use instead? I'm thinking of using explained variance (R-squared).
1
-
$\begingroup$ You might see the responses and discussion here: stats.stackexchange.com/questions/49010/… $\endgroup$– itpetersenCommented Feb 10, 2021 at 14:01
Add a comment
|