Figure 1From: Computational identification of surrogate genes for prostate cancer phases using machine learning and molecular network analysisIFS curves determining the number of features. We used the IFS curves to determine the number of features used in mRMR computation. “Prediction 1”, “Prediction 2” and “Prediction 3” refer to the three prediction approaches, a vote of the top five nearest neighbors, the first nearest neighbor and nearest clustering center of each phenotype group, seperately. It is noted that when using four genes, the prediction accuracy reaches the maximum.Back to article page