Skip to main content

Advertisement

Fig. 2 | BMC Bioinformatics

Fig. 2

From: GARS: Genetic Algorithm for the identification of a Robust Subset of features in high-dimensional datasets

Fig. 2

Radar plots that summarize the performance of the different algorithms tested in a ‘binary low-dimension dataset’. To test the efficacy of each algorithm, we calculated ACC = Accuracy, SEN = Sensitivity, SPE = Specificity, PPV = Positive Predictive Value, NPV = Negative Predictive Value, AUC = Area Under ROC Curve, and Nfeats = n. of selected features on the independent test set. To evaluate the efficiency of each algorithm, we measured the average learning time for each cross-validation fold (Time). To get an overall assessment of the algorithm performance, we calculated the area of the polygon obtained connecting each point of the aforementioned measurements: the wider the area, the better the overall performance. GARS (red chart) and LASSO (purple chart) covered 98% of the total area, SBF (green chart) 91%, rfGA (yellow chart) 87%, svmGA (light blue chart) 76% and RFE (blue chart) 70%

Back to article page