Nonlinear Multivariate Calibration Methods in Analytical Chemistry

Sekulic, Sonja; Seasholtz, Mary Beth; Wang, Ziyi; Kowalski, Bruce R.; Lee, Samuel E.; Holt, Bradley R.

doi:10.1021/ac00067a714

Cited by 102 publications

(32 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In our study, non-linear relationships between spectral and chemical data may not have occurred due to the smaller range of CP values, and consequently no improvements were found with the use of the local regression. Sekulic et al (1993) state that non-linear methods such as local regression are able to construct models with equivalent predictive ability compared with PLS regression or principal component regression when the relationship between predicted v. measured values is linear.…”

Section: Nir Spectral Regionsmentioning

confidence: 99%

NIRS prediction of the feed value of temperate forages: efficacy of four calibration strategies

Andueza

Picard

Jestin

et al. 2011

Animal

View full text Add to dashboard Cite

Near infrared reflectance spectroscopy (NIRS) of 924 fresh temperate forages were used to develop calibration models for chemical composition -crude ash (CA) and crude protein (CP) -organic matter digestibility (OMD) and voluntary intake (VI). We used 110 samples to assess the models. Four calibration strategies for determining forage quality were compared: (i) speciesspecific calibration, (ii) family-specific calibration, (iii) a global procedure and (iv) a local approach. Forage calibration data sets displayed CA values ranging from 52 to 205 g/kg of dry matter (DM), CP values from 50 to 280 g/kg DM, OMD values from 0.48 to 0.85 g/g and VI values from 22.5 to 115.2 g DM/kg metabolic body weight (BW 0.75 ). The calibration models performed well for all the variables except for VI. For CA, local procedure showed lower standard error of prediction (SEP) than species-specific, family-specific or global models. For CP, the calibration models all showed similar SEP values (11.13, 11.08, 11.38 and 11.34 g/kg DM for species-specific, family-specific, global and local approaches). For OMD, the local procedure gave a similar SEP (0.024 g/g) to specific species and global procedures (0.027 g/g) and a lower SEP than the family-specific approach (0.028 g/g). For VI, the local approach and species-specific calibration showed lower SEP (7.08 and 7.16 g/kg BW 0.75 ) than the broad-based calibrations (8.09 and 8.34 g/kg BW 0.75 for family-specific model and global procedure, respectively). Local calibration may thus offer a practical way to develop robust universal equations for animal response determinations.Keywords: chemical composition, in vivo organic matter digestibility, voluntary intake, near infrared reflectance spectroscopy ImplicationsThis study evaluates the utility of different near infrared reflectance spectroscopy (NIRS) calibration methods for predicting the feed value of fresh forages. We compare an alternative local approach with the more classical global approach. The local approach conserves all the advantages of NIRS technology (rapidity, non-contaminant technology and precision) and takes into account the number of samples and their diversity when large database are available. It overcomes the main limitation of NIRS, which is that the samples for prediction have to be close to those of the calibration population. The local approach could improve prediction of the nutritional value of feeds, and thus have important economic and environmental implications for animal nutrition. IntroductionRuminant performance is closely related to the digestibility and intake of feedstuffs. In most countries, knowledge of the feed value of forages is based on in vivo measurements made in castrated male sheep (Demarquilly et al., 1995 for a review). However, the large-scale application of in vivo digestibility trials for evaluating forage feed value is an expensive and time-consuming process that requires large amounts of feeds. As knowing these values is essential for controlled animal feeding, several methods have been d...

show abstract

Section: Nir Spectral Regionsmentioning

confidence: 99%

NIRS prediction of the feed value of temperate forages: efficacy of four calibration strategies

Andueza

Picard

Jestin

et al. 2011

Animal

View full text Add to dashboard Cite

show abstract

“…The number of descriptor or independent variables and predictor or dependent variables determine the number of nodes in input and output layers respectively. 27,28 Principal component-artificial neural networks (PC-ANNs) have good speed and low calibration errors, because of reducing the input data for ANN. 29 In this type of networks, principal component analysis was first applied to data and then the scores on the selected principal components were subjected to ANN as input.…”

Section: Pls and Artificial Neural Networkmentioning

confidence: 99%

Comparison of Artificial Neural Networks with Partial Least Squares Regression for Simultaneous Determinations by ICP‐AES

et al. 2007

View full text Add to dashboard Cite

Simultaneous determination of several elements (U, Ta, Mn, Zr and W) with inductively coupled plasma atomic emission spectrometry (ICP-AES) in the presence of spectral interference was performed using chemometrics methods. True comparison between artificial neural network (ANN) and partial least squares regression (PLS) for simultaneous determination in different degrees of overlap was investigated. The emission spectra were recorded at uranium analytical line (263.553 nm) with a 0.06 nm spectral window by ICP-AES. Principal component analysis was applied to data and scores on 5 dominant principal components were subjected to ANN. A 5-5-5 (input, hidden and output neurons) network was used with linear transfer function after both hidden and output layers. The PLS model was trained with five latent variables and 20 samples in calibration set. The relative errors of predictions (REP) in test set were 3.75% and 3.56% for ANN and PLS respectively.

show abstract

“…They can approximate non-linear cases, but they will need more components for that. Many non-linear approaches have been proposed 6,30 . However, often many model parameters have to be estimated.…”

Section: Partial Least Squares (Pls)mentioning

confidence: 99%

Online Monitoring of Gravity, FAN and β-Glucan during Mashing

Mitzscherling

Becker

Delgado³

et al. 2007

Journal of the Institute of Brewing

View full text Add to dashboard Cite

The paper presents a measurement system that is based on an array of non-specific sensors and the installation of both in a pilot and in an industrial mash tun. Three calibration strategies were pursued to predict gravity, FAN and ␤-glucan during mashing in combination with three different multivariate calibrations (PCR, CPCR, PLS). The over all best calibration had RMSPEs of 0.78%, 1.56 mg/100 mL and 181.73 mg/L for gravity, FAN and ␤-glucan, respectively. The measurement system makes it possible to monitor online the three major sub-processes of mashing (amylolysis, proteolysis, and cytolysis), which is the precondition for a situational control of the mashing process.

show abstract

Nonlinear Multivariate Calibration Methods in Analytical Chemistry

Abstract: Over the years the role of the analytical chemist has evolved in step with technological advancements and the quest to explore and understand issues of increasing complexity. The

Cited by 102 publications

References 43 publications

NIRS prediction of the feed value of temperate forages: efficacy of four calibration strategies

NIRS prediction of the feed value of temperate forages: efficacy of four calibration strategies

Comparison of Artificial Neural Networks with Partial Least Squares Regression for Simultaneous Determinations by ICP‐AES

Online Monitoring of Gravity, FAN and β-Glucan during Mashing

Contact Info

Product

Resources

About