Figure 2
From: The difficulties of reproducing conventionally derived results through 500k-chip technology
Optimization of neural network classifiers. Iterative optimization of the starting configuration by systematically adding/removing genomic loci while fitting the NN model to the set of 1,042 observations under the constraint of reproducibility with ten-fold cross-validation. The red circles designate the percentage of correctly classified subjects for each optimization step, with optimization steps being plotted along the x-axis (over-proportionally large drops in performance indicate removal of loci of larger weight).