Figure 5.

Quintile rank of prediction accuracy by sample size and analysis algorithm for Simulated_Large, Simulated_Small, Imaging_Small, and Imaging_Large. Shown ranks are the quintile into which the median prediction accuracy for each method within each data type and cell fell across the distribution of all analysis iteration for each data type and cell. RF: Random Forest; MR: Multiple Regression; EN: Elastic Net; MKL: Multiple Kernel Learning; KRR: Kernel Ridge Regression; GPR: Gaussian Process Regression. Color bars and plot coloring show the rank from zero to five, with higher values (warmer colors) indicating higher rank and therefore better prediction accuracy.