Limits...
Optimization of Parameter Selection for Partial Least Squares Model Development.

Zhao N, Wu ZS, Zhang Q, Shi XY, Ma Q, Qiao YJ - Sci Rep (2015)

Bottom Line: In multivariate calibration using a spectral dataset, it is difficult to optimize nonsystematic parameters in a quantitative model, i.e., spectral pretreatment, latent factors and variable selection.In this study, we describe a novel and systematic approach that uses a processing trajectory to select three parameters including different spectral pretreatments, variable importance in the projection (VIP) for variable selection and latent factors in the Partial Least-Square (PLS) model.The PLS model optimizes modeling parameters step-by-step, but the robust model described here demonstrates better efficiency than other published papers.

View Article: PubMed Central - PubMed

Affiliation: 1] Beijing University of Chinese Medicine, Beijing 100102, China [2] Beijing Key Laboratory for Basic and Development Research on Chinese Medicine, Beijing, 100102, China [3] Key Laboratory of TCM-information Engineer of State Administration of TCM, Beijing, China, 100102.

ABSTRACT
In multivariate calibration using a spectral dataset, it is difficult to optimize nonsystematic parameters in a quantitative model, i.e., spectral pretreatment, latent factors and variable selection. In this study, we describe a novel and systematic approach that uses a processing trajectory to select three parameters including different spectral pretreatments, variable importance in the projection (VIP) for variable selection and latent factors in the Partial Least-Square (PLS) model. The root mean square errors of calibration (RMSEC), the root mean square errors of prediction (RMSEP), the ratio of standard error of prediction to standard deviation (RPD), and the determination coefficient of calibration (Rcal(2)) and validation (Rpre(2)) were simultaneously assessed to optimize the best modeling path. We used three different near-infrared (NIR) datasets, which illustrated that there was more than one modeling path to ensure good modeling. The PLS model optimizes modeling parameters step-by-step, but the robust model described here demonstrates better efficiency than other published papers.

No MeSH data available.


Related in: MedlinePlus

Schematic diagram of processing trajectory and assessment of PLS model corn samples (a), Yinhuang granules samples (b) and pharmaceutical tablets sample (c).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4499800&req=5

f2: Schematic diagram of processing trajectory and assessment of PLS model corn samples (a), Yinhuang granules samples (b) and pharmaceutical tablets sample (c).

Mentions: Using the corn dataset as an example, the calibration spectra were preprocessed with different methods including Standard Normal Variate (SNV) and Savitzky-Golay smoothing with 9 points (SG(9)), as well as SG(9) combined with the derivative spectra. The latent factor was set from 1 to 10 to avoid over-fitting. VIP was then used to select variables with different latent factors. Finally, the process routines from PLS model development and validation were selected (Fig. 2). The parameters for PLS models in water, baicalin and API are shown in Table. 1s-3s. There are various trends in the model evaluation indexes. In Fig. 2a, we see that the RMSEC and RMSEP decreased and the Rcal2, Rpre2, and RPD increased with increasing latent factor coupled with different pretreatment methods.


Optimization of Parameter Selection for Partial Least Squares Model Development.

Zhao N, Wu ZS, Zhang Q, Shi XY, Ma Q, Qiao YJ - Sci Rep (2015)

Schematic diagram of processing trajectory and assessment of PLS model corn samples (a), Yinhuang granules samples (b) and pharmaceutical tablets sample (c).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4499800&req=5

f2: Schematic diagram of processing trajectory and assessment of PLS model corn samples (a), Yinhuang granules samples (b) and pharmaceutical tablets sample (c).
Mentions: Using the corn dataset as an example, the calibration spectra were preprocessed with different methods including Standard Normal Variate (SNV) and Savitzky-Golay smoothing with 9 points (SG(9)), as well as SG(9) combined with the derivative spectra. The latent factor was set from 1 to 10 to avoid over-fitting. VIP was then used to select variables with different latent factors. Finally, the process routines from PLS model development and validation were selected (Fig. 2). The parameters for PLS models in water, baicalin and API are shown in Table. 1s-3s. There are various trends in the model evaluation indexes. In Fig. 2a, we see that the RMSEC and RMSEP decreased and the Rcal2, Rpre2, and RPD increased with increasing latent factor coupled with different pretreatment methods.

Bottom Line: In multivariate calibration using a spectral dataset, it is difficult to optimize nonsystematic parameters in a quantitative model, i.e., spectral pretreatment, latent factors and variable selection.In this study, we describe a novel and systematic approach that uses a processing trajectory to select three parameters including different spectral pretreatments, variable importance in the projection (VIP) for variable selection and latent factors in the Partial Least-Square (PLS) model.The PLS model optimizes modeling parameters step-by-step, but the robust model described here demonstrates better efficiency than other published papers.

View Article: PubMed Central - PubMed

Affiliation: 1] Beijing University of Chinese Medicine, Beijing 100102, China [2] Beijing Key Laboratory for Basic and Development Research on Chinese Medicine, Beijing, 100102, China [3] Key Laboratory of TCM-information Engineer of State Administration of TCM, Beijing, China, 100102.

ABSTRACT
In multivariate calibration using a spectral dataset, it is difficult to optimize nonsystematic parameters in a quantitative model, i.e., spectral pretreatment, latent factors and variable selection. In this study, we describe a novel and systematic approach that uses a processing trajectory to select three parameters including different spectral pretreatments, variable importance in the projection (VIP) for variable selection and latent factors in the Partial Least-Square (PLS) model. The root mean square errors of calibration (RMSEC), the root mean square errors of prediction (RMSEP), the ratio of standard error of prediction to standard deviation (RPD), and the determination coefficient of calibration (Rcal(2)) and validation (Rpre(2)) were simultaneously assessed to optimize the best modeling path. We used three different near-infrared (NIR) datasets, which illustrated that there was more than one modeling path to ensure good modeling. The PLS model optimizes modeling parameters step-by-step, but the robust model described here demonstrates better efficiency than other published papers.

No MeSH data available.


Related in: MedlinePlus