Skip to main content
. 2012 Mar;19(3):251–260. doi: 10.1089/cmb.2011.0078

FIG. 1.

FIG. 1.

The absolute values of regression coefficients of the 252 Z-curve variables in the PLS model (fragments with 192-bp length of dataset-2). In order to identify key variables that characterize different patterns, the contribution of each variable to pattern recognition models is necessary to estimate quantitatively. And for univariate regression problems, the regression coefficient of each variable is the reasonable quantitative measurement. According to the results, it is clear that only a few variables' coefficients are remarkably larger than other variables'.