Models have been built using multiple learning algorithms including support vector machine and random forest.
The prediction models were built using Multiple Linear Regression (MLR) as the fitting method, coupled with the Genetic Algorithm (GA) as variable subset selection strategy and the statistical parameter Q loo 2 ("leave-one-out" cross validation) as the fitness function.
Variable selection was based on the genetic algorithm (GA) and ordered predictor selection (OPS) approaches, and the models were built using the multiple linear regression (MLR) and partial least squares (PLS) regression methods.
Logistic regression models were built using the multiple fractional polynomial (MFP) model selection procedures available in the MFP package (Royston and Altman, 1994), following the fractional polynomial model selection procedures using each protein individually.
The relationship between the change multiple of porosity and the scour pore volume multiple is built using a secondary variation pattern. 5.
The relationship between the change multiple of porosity and the scour pore volume multiple is built using a secondary variation pattern.
Multiple models were built using different number of principle components, and we chose 4 principle components as the discriminants and LDA as the classification method that combined to minimize the error rate in leave-one-out cross validation (LOOCV).
Multiple antigen classifiers were built using Support Vector Machines (SVMs).
Based on a statistical model in a previous study (Equation 5 in [16]), a multiple regression model was built using SPSS 15 to test the slope and Ct differences between equimolar circular and linear DNA in each qPCR.
Multiple sequence alignments were built using the ClustalW tool.
