摘要
中药指纹图谱数据具有变量数很大而样本数较小的特点,本文中采用拉格朗日求极值的方法导出一种新的适合用于处理这类数据的偏最小二乘算法。结果表明:所得到新的算法,在处理中药指纹图谱数据时,与传统的偏最小二乘算法比较,节省存储单元,计算量小,计算速度快,因而计算效率高。
The fingerprinting data sets of Chinese medicine are the data sets with large number of variables and few objects. A new kind of PLS algorithm that is suitable to deal with this kind of data sets has been derived by use of the Lagrange method of solving extremum problem in this paper. The results indicate that by comparing with the traditional PLS algorithm the new presented algorithm is memory-saving, with small amount of calculation, fast and effective when dealing with the fingerprinting data of Chinese medicine or the data with large number of variables and few objects.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2005年第8期639-642,共4页
Computers and Applied Chemistry
基金
福建省自然科学基金(C0210006)资助项目
关键词
偏最小二乘法
指纹图谱
回归分析
partial least squares, fingerprinting, regression analysis
