Figure 3.

Five-fold cross-validation performance of models trained using amino acid composition with varying window lengths. To determine which window sizes can be used for generating the predictive model that best identifies carboxylation sites, a five-fold cross-validation is conducted to evaluate the models trained with different window lengths 2n+1, where n changes from four to ten.

Lee et al. BMC Bioinformatics 2011 12(Suppl 13):S10   doi:10.1186/1471-2105-12-S13-S10