For the two seemingly unrelated regression system, this paper proposed a new type of estimator called pre-test principal components estimator (PTPCE) and discussed some properties of PTPCE.
将低场核磁共振(low field nuclear magnetic resonance,LF-NMR)分析技术应用于煎炸油脂总极性化合物(total polar compounds,TPC)含量的预测。采用柱层析方法测定油脂样品的TPC含量作为测定值,采集油脂样品的LF-NMR弛豫特性(峰起始时间...将低场核磁共振(low field nuclear magnetic resonance,LF-NMR)分析技术应用于煎炸油脂总极性化合物(total polar compounds,TPC)含量的预测。采用柱层析方法测定油脂样品的TPC含量作为测定值,采集油脂样品的LF-NMR弛豫特性(峰起始时间T21、T22、T23相应的峰面积比例S21、S22、S23、单组分弛豫时间T2W),分别利用向后筛选多元回归分析、主成分回归分析和偏最小二乘回归分析建立LF-NMR弛豫特性与TPC含量的回归方程,比较3种模型的校正集和预测集的决定系数与均方根误差,最终确定最优模型为偏最小二乘回归模型。应用此模型预测预测集样品TPC含量,决定系数R2可达0.928,预测集均方根误差为0.568%,模型稳定。展开更多
文摘For the two seemingly unrelated regression system, this paper proposed a new type of estimator called pre-test principal components estimator (PTPCE) and discussed some properties of PTPCE.
文摘将低场核磁共振(low field nuclear magnetic resonance,LF-NMR)分析技术应用于煎炸油脂总极性化合物(total polar compounds,TPC)含量的预测。采用柱层析方法测定油脂样品的TPC含量作为测定值,采集油脂样品的LF-NMR弛豫特性(峰起始时间T21、T22、T23相应的峰面积比例S21、S22、S23、单组分弛豫时间T2W),分别利用向后筛选多元回归分析、主成分回归分析和偏最小二乘回归分析建立LF-NMR弛豫特性与TPC含量的回归方程,比较3种模型的校正集和预测集的决定系数与均方根误差,最终确定最优模型为偏最小二乘回归模型。应用此模型预测预测集样品TPC含量,决定系数R2可达0.928,预测集均方根误差为0.568%,模型稳定。
基金partly supported by the National Natural Science Foundation of China(Grant No.11271039)Education Ministry Funds for Doctor Supervisors and Fund from Collaborative Innovation Center on Capital Social Construction and Social Management(Grant No.006000546615539)supported by the National Natural Science Foundation of China(Grant No.11501018)
