causal

Complex Models Fit Past But Fail to Predict

Complex methods work best at explaining the past, but under uncertainty where prediction is needed you must scale back to simpler tools; a fast-and-frugal tree using three factors can outperform logistic regression in predicting heart-patient outcomes going forward, while also being understandable to doctors and adaptable when the patient population changes.

causalpending

Speaker

Gerd Gigerenzer

Evidence Quote

“under uncertainty, predicting the future... make it simple. If you want to explain the past, make it complicated.”

Source

Gerd Gigerenzer on Gut Feelings— EconTalk
Created: 6/13/2026, 12:10:34 AM

Your notes

Loading notes...