Table 3.

Performance of a linear model by adding flexibility to assumed linear association with the outcome

Variable	R-squared*
Predictor linear	0.00938
Predictor with splines with one knot	0.9853
Predictor with fractional polynomial	0.9992

*R-squared is measure of how close the model fits the data, 1 indicates the model explains all the variability of the data, whereas with 0 the model does not explain any variability. For other types of models similar measurements are available