Figure 2.

Classification performance of different types of classifiers. The performance of the best SVM in each category is compared to three other standard machine learning methods: LDA (Linear Discriminant Analysis), Decision Tree, and ANN (Artificial Neural Networks) and a simple fold-change-based predictor. Performance of machine learning methods is evaluated and accuracy is measured using LOOCV. Labelling of panels is as follows, "microarray": using genome-wide mRNA microarray profiling data; "chip": using genome-wide ChIP-seq of transcription factors data; "micro-chip": using both microarray and ChIP-seq. The fold-change-based predictor results are only under the "microarray" panel since it uses only microarray data.

Xu et al. BMC Systems Biology 2010 4:173   doi:10.1186/1752-0509-4-173
Download authors' original image