The model in which a partial least

使用偏最小二乘回归算法的模型比基于主成分回归的模型效果更好。测试样品中双氯芬酸钠测定的误差分别等于2.2和2.6％。对于PLS模型，添加剂的RSEP值在0.4-2.4％的范围内变化，对于PCR而言，其RSEP值在0.5-2.1％的范围内变化。<br><br>在建立氨茶碱多变量模型的过程中，应用了两个光谱范围：3570-2587和1716-415 cm-1。茶碱的校准曲线的特征是R2值在0.983–0.999之间（图5）。在测试样品的情况下，茶碱的RSEP值对于PLS等于2.7％，对于PCR方法等于2.3％。表2收集了API的全套校准结果。正如人们可能会注意到的那样，多元模型的RSEP值甚至比研究的两种API的单变量方法发现的值小几倍。

The model in which a partial least squares regression algorithm was used worked slightly better than the one based on principal component regression. The errors found for diclofenac sodium determination in testing samples equal 2.2 and 2.6%, respectively. RSEP values for additives varied in the 0.4–2.4% range for the PLS model and in the 0.5–2.1% range for the PCR one.<br><br>In the course of constructing multivariate models for aminophylline, two spectral ranges were applied: 3570–2587 and 1716–415 cm−1. The calibration curves for theophylline were characterised by R2 values in the range 0.983–0.999 (Fig. 5). The RSEP values obtained for theophylline in the case of the testing sample set equal 2.7% for PLS and 2.3% for PCR methods. A full set of calibration results for the API is collected in Table 2. As one could notice, the RSEP values for the multivariate models were even a few times smaller than those found for the univariate approach for both the studied APIs.

使用偏最小二乘回归算法的模型比基于主成分回归的模型工作得稍好。双氯芬酸钠测定样品的误差分别为2.2%和2.6%。对于PLS模型，添加剂的RSEP值在0.4–2.4%范围内变化，而对于PCR模型，RSEP值在0.5–2.1%范围内变化。<br>在构建氨茶碱多变量模型的过程中，应用了两个光谱范围：3570-2587和1716-415cm-1。茶碱的校准曲线由0.983–0.999范围内的R2值表征（图5）。在测试样本集的情况下，茶碱的RSEP值对于PLS为2.7%，对于PCR方法为2.3%。表2收集了API的全套校准结果。正如我们可以注意到的，多变量模型的RSEP值甚至比所研究的两个api的单变量方法的RSEP值小几倍。