Multidimensional correlated variables were converted into low dimensional independent main-eigenvector by means of singularity value decomposition.

• 奇异值分解可以将高维相关变量压缩为低维独立的主特征矢量,而支持向量机可以完成模式识别和非线性回归。