Fig. 4
From: The roles of predictors in cardiovascular risk models - a question of modeling culture?
Partial dependence of estimated risk on a age, b total cholesterol, c blood glucose, d BMI, showing how average predictions vary with these variables when keeping fixed all other predictors. Black: overall; red: at age fixed at 40 years and sex set to female; yellow: 50 years, female; green: 60 years, female; blue: 70 years, female. Dotted lines: single-layer neural networks/logistic regression; dashed-dotted lines: generalized additive models; dashed lines: multilayer neural networks; solid lines: extreme gradient boosted trees. All models were fitted at full data availability