用于分类的主判别变量算法(principal discriminant variate algorithm) - principal discriminant variate algorithm for two classes, x1 and x2.



x1 is the training samples in the 1st class with each column being the spectrum characterizing the corresponding sample.



x2 is the training samples in the 2nd class with each column being the spectrum characterizing the corresponding sample.



lamda is the penalty weight defining the balance between FLDA and PCA.



dvnum is the number of principal discriminant variates to be computed. A defaul value for dvnum is 2.



dv is the principal discriminant variates.



scores is the matrix with each row being the scores of the samples on the corresponding principal discriminant variate.



xmean is the mean spectra of the sample, which was used for mean-centering the data in training and prediction.