Quantitative model assessment method and system based on molecular spectrum analysis

Abstract

The invention discloses a quantitative model assessment method and system based on molecular spectrum analysis. The quantitative model assessment method based on the molecular spectrum analysis comprises the following steps that: carrying out assessment on the sample amount of a quantitative model; carrying out assessment on the model accuracy of the quantitative model; carrying out assessment on the model correlation of the quantitative model; carrying out assessment on the model representativeness of the quantitative model; and on the basis of the assessment results of the sample amount, the model accuracy and the model representativeness, carrying out assessment on the quantitative model, and analyzing an assessment result to finish the molecular spectrum analysis. The quantitative model assessment system based on the molecular spectrum analysis comprises a sample amount assessment module, a model accuracy assessment module, a model correlation assessment module, a model representativeness assessment module, a quantitative model scoring module and a scoring result analysis module. By use of the method and the device, the advantages that an analysis result is simply and quickly obtained and the comprehensiveness of the assessment method can be comprehensively considered are realized.

Description

technical field [0001] The invention relates to the field of molecular spectrum analysis, in particular to a quantitative model evaluation method and system based on molecular spectrum analysis. Background technique [0002] At present, most of the methods of spectroscopy combined with chemometrics for quantitative analysis adopt the same mode, that is, to establish a calibration model based on a set of known samples. This group of known samples is called the calibration set samples, through the spectra of this group of samples and their corresponding basic data (physical and chemical data), the calibration model is established by using the multivariate correction method. For the sample to be tested, it only needs to measure its spectrum, and the quantitative result can be given quickly according to the established model. [0003] In the process of model establishment and verification, some statistical parameters will be used as model evaluation parameters, such as deviatio...

AI Technical Summary

Problems solved by technology

[0005] In terms of molecular spectroscopic analysis technology, there is a problem of complex evaluation due to different measurement scales and data dimension differences between samples, which will affect the evaluation parameters of commonly used models.

[0006] Moreover, 1. The evaluation parameters in the existing methods cannot reflect the influence of the number of samples in the calibration set on the model during modeling, and the quality of the model cannot be well observed only by using the commonly used SEP and RP values. A small sample set can also get a good SEP and RP evaluation for model creation

2. SEP will be affected by the physical and chemical data of the calibration set. The model built by the sample set with large physical and chemical values ​​has a relatively large SEP, which leads to the incomparability of SEP between different sample models

4. The evaluation parameters of various models are independent of each other, and it is impossible to conduct a more comprehensive comprehensive evaluation of the model

It is difficult for those who are not familiar with chemometrics to easily know the quality of the model through the existing evaluation parameters

Images