Comparison among principal component, partial least squares and back propagation neural network analyses for accuracy of measurement of selected soil properties with visible and near infrared spectroscopy

Mouazen, Abdul Mounem; Kuang, Boyan; Baerdemaeker, Josse De; Ramón, Herman

doi:10.1016/j.geoderma.2010.03.001

Cited by 381 publications

(264 citation statements)

References 33 publications

Supporting

Mentioning

242

Contrasting

Unclassified

Order By: Relevance

“…Comparisons of the reflectance-based models using Scheme 7 and Scheme 8 are shown in Figure 6c,d. The former has a smaller absolute prediction bias and RPD, whereas the latter has an RPD value of greater than 2.0, which indicates a very good result [43]. The VIP scores of this model are shown in Figure 7.…”

Section: Calibration and Validationmentioning

confidence: 91%

“…Residual prediction deviation (RPD), along with the coefficient of determination (r 2 ) and the RMSE for the cross-validation (r 2 cv , RMSECV) and prediction (r 2 Pre , RMSEP), were computed to interpret the model predictive ability [42,43]. For the VNIR estimation of the soil properties, RPD > 1.4 indicates an acceptable predictive ability for the model [43].…”

Section: Plsr Modeling Of Soil Organic Mattermentioning

confidence: 99%

See 1 more Smart Citation

Transferability of a Visible and Near-Infrared Model for Soil Organic Matter Estimation in Riparian Landscapes

et al. 2014

View full text Add to dashboard Cite

Abstract:The transferability of a visible and near-infrared (VNIR) model for soil organic matter (SOM) estimation in riparian landscapes is explored. The results indicate that for the soil samples with air-drying, grinding and 2-mm sieving pretreatment, the model calibrated from the soil sample set with mixed land-use types can be applied in the SOM prediction of cropland soil samples (r 2 Pre = 0.66, RMSE = 2.78 g•kg −1 , residual prediction deviation (RPD) = 1.45). The models calibrated from cropland soil samples, however, cannot be transferred to the SOM prediction of soil samples with diverse land-use types and different SOM ranges. Wavelengths in the region of 350-800 nm and around 1900 nm are important for SOM estimation. The correlation analysis reveals that the spectral wavelengths from the soil samples with and without the air-drying, grinding and 2-mm sieving pretreatment are not linearly correlated at each wavelength in the region of 350-1000 nm, which is an important spectral region for SOM estimation in riparian landscapes. This result explains why the models calibrated from samples without pretreatment fail in the SOM estimation. The Kennard-Stone algorithm performed well in the selection of a representative subset for SOM estimation using the spectra of soil samples with pretreatment, but failed in soil samples without the pretreatment. Our study also demonstrates that a widely applicable SOM prediction model for riparian landscapes should be based on a wide range of SOM content.

show abstract

Section: Calibration and Validationmentioning

confidence: 91%

Section: Plsr Modeling Of Soil Organic Mattermentioning

confidence: 99%

Transferability of a Visible and Near-Infrared Model for Soil Organic Matter Estimation in Riparian Landscapes

et al. 2014

View full text Add to dashboard Cite

show abstract

“…This method is similar to the approach in previous studies for the determination of the optimal input number in PLS (Liu et al, 2010) and PCR (Serneels and Verdonck, 2009). However, this method contrasts with the approach that based on the cumulative percentage of explained data variance (Mouazen et al, 2010). Although the determination of the optimal PCs number still remains as an open question (Abdi and Williams, 2010), PCs with small variance (less than 0.1% variance) should not be ignored without proper investigation and evaluation.…”

Section: Optimal Number Of Principal Componentsmentioning

confidence: 99%

“…Recently, Mouazen et al (2010) advocated that ANN with latent variables (LVs) was the best when compared with PCs-ANN, PCR, and PLS models in soil property assessments. However, this is worth to highlight that latent variables are generally not recommended to be the inputs of ANN.…”

Section: Artificial Neural Network With Principal Components (Pcs-ann)mentioning

confidence: 99%

Neural network and principal component regression in non-destructive soluble solids content assessment: a comparison

Chia

Rahim

2012

J. Zhejiang Univ. Sci. B

View full text Add to dashboard Cite

Abstract:Visible and near infrared spectroscopy is a non-destructive, green, and rapid technology that can be utilized to estimate the components of interest without conditioning it, as compared with classical analytical methods. The objective of this paper is to compare the performance of artificial neural network (ANN) (a nonlinear model) and principal component regression (PCR) (a linear model) based on visible and shortwave near infrared (VIS-SWNIR) (400-1000 nm) spectra in the non-destructive soluble solids content measurement of an apple. First, we used multiplicative scattering correction to pre-process the spectral data. Second, PCR was applied to estimate the optimal number of input variables. Third, the input variables with an optimal amount were used as the inputs of both multiple linear regression and ANN models. The initial weights and the number of hidden neurons were adjusted to optimize the performance of ANN. Findings suggest that the predictive performance of ANN with two hidden neurons outperforms that of PCR.

show abstract

“…In order to improve the accuracy of SOC calibration models with VIS/NIR spectral data, the raw spectral data are often pre-processed before modeling [9,32,33]. Pre-processing is usually regarded as an integral part of chemometrics modeling with spectral data.…”

Section: Introductionmentioning

confidence: 99%

Estimating Soil Organic Carbon Using VIS/NIR Spectroscopy with SVMR and SPA Methods

et al. 2014

View full text Add to dashboard Cite

With 298 heterogeneous soil samples from Yixing (Jiangsu Province), Zhongxiang and Honghu (Hubei Province), this study aimed to combine a successive projections algorithm (SPA) with a support vector machine regression (SVMR) model (SPA-SVMR model) to improve the estimation accuracy of soil organic carbon (SOC) contents using the laboratory-based visible and near-infrared (VIS/NIR, 350−2500 nm) spectroscopy of soils. The effects of eight spectra pre-processing methods, i.e., Log (1/R), Log (1/R) coupled with Savitzky-Golay (SG) smoothing (Log (1/R) + SG), first derivative with SG smoothing (FD), second derivative with SG smoothing (SD), SG, standard normal variate (SNV), mean center (MC) and multiplicative scatter correction (MSC), on SPA-based informative wavelength selection were explored. The SVMR model (i.e., SVMR without SPA) and SPA-PLSR model (i.e., SPA combined with partial least squares regression (PLSR)) were developed and compared with the SPA-SVMR model in order to evaluate the performance of SPA-SVMR. The results indicated that the variables selected by SPA and their distributions were strongly affected by different pre-processing methods, and SG was the optimal pre-processing method for SPA-SVMR model development; the SPA-SVMR model using SG pre-processing and 28 SPA-selected wavelengths obtained a better result (R 2 V = 0.73, RMSE V = 2.78 g·kg and RPD V = 1.63). Most of the spectral bands used by the SPA-SVMR model over the near-infrared region were important wavelengths for SOC content estimation. This study demonstrated that the combination of SPA and SVMR is feasible and reliable for estimating SOC content from the VIS/NIR spectra of soils in regions with multiple soil and land-use types.

show abstract

Comparison among principal component, partial least squares and back propagation neural network analyses for accuracy of measurement of selected soil properties with visible and near infrared spectroscopy

Cited by 381 publications

References 33 publications

Transferability of a Visible and Near-Infrared Model for Soil Organic Matter Estimation in Riparian Landscapes

Transferability of a Visible and Near-Infrared Model for Soil Organic Matter Estimation in Riparian Landscapes

Neural network and principal component regression in non-destructive soluble solids content assessment: a comparison

Estimating Soil Organic Carbon Using VIS/NIR Spectroscopy with SVMR and SPA Methods

Contact Info

Product

Resources

About