Figure 4
From: Effects of sample size on robustness and prediction accuracy of a prognostic gene signature
Concordance between predicted outcomes increases with an increasing training sample size. For each sample size from 100 to 500 by increments of 100, one hundred samples were first selected as testing samples and 100 independently selected training samples were used to predict the outcomes of the already selected testing samples. Concordance of outcome prediction between each pair of 100 predictions (a total of 4950 pairs) was calculated. Three different algorithms (A. DLDA, B, RF, and C. SVM) were tested.