Figure 3
From: Feature selection for splice site prediction: A new method using EDA-based feature ranking
Visualization of EDA-R feature weights for acceptor prediction. For each of the three datasets, the color coded feature weights as a result of the EDA-R feature selection in combination with a linear SVM are shown. (A) The simplest dataset (only position dependent nucleotides, 400 binary features). (B) The extended (position dependent nucleotides + position invariant 3-mers, 528 binary features). (C) The most complex dataset (also including position dependent dinucleotides, 2096 binary features).