Red Door Analytics

Building and validating a model that gives an individual their risk. Discrimination says whether the ordering is right; calibration says whether the numbers are, and a model can get the first right while getting the second badly wrong.

Also called risk prediction, prognostic models.

Prediction and risk modelling: Predicted risk against what was observed, with the distribution of predictions beneath.
Predicted risk against what was observed, with the distribution of predictions beneath. Discrimination says whether the ordering is right; calibration says whether the numbers are. A model can rank every pair correctly and still tell people the wrong risk, and the c-statistic will not move.

Working on prediction and risk modelling?

We do this every week — as a collaborator on your study, or as the people who wrote the software you are using.

Get in touch Start a project