A typical example in simulation I. This graph presents the average L2 risk of the final prediction model on the candidate list from the UR-VIM and the TMLE-VIM, for simulation I data with setup (σe = 5, mw = 10, ρ = 0.7). In the left panel, the MVR risk is plotted against a series of p-value thresholds used to truncate the candidate list; the right panel plots the D/S/A risk. The testing set (a) predictions are grouped in solid blue lines, and the testing set (b) predictions are grouped in broken orange lines. Dots represent the UR-VIM, and triangles represent the TMLE-VIM.
Wang and van der Laan BMC Bioinformatics 2011 12:312 doi:10.1186/1471-2105-12-312