Fig 1.
Histogram distribution of Cr content for the 165 soil samples.
Fig 2.
The FT-NIR spectra of 165 soil samples.
(a) The raw data (b) The SG-pretreated data.
Table 1.
Descriptive statistics of Cr contents for the training and testing sample sets, respectively.
Fig 3.
The flowchart of BDE-PSSVM modeling procedure.
Fig 4.
The grid search prediction results by parametric scaling of for SVM model.
Fig 5.
The iterative trend of BDE optimization in combination with PSSVM parametric scaling.
Fig 6.
The BDE iterative results by tuning the specific parameters of {NP, L, CT}.
(a) NP = 1 (b) NP = 2 (c) NP = 3 (d) NP = 4 (e) NP = 5 (f) NP = 6 (g) NP = 7 (h) NP = 8.
Fig 7.
The 56 feature variables selected by BDE-PSSVM.
Table 2.
Comparison of the optimal models by different methods with parametric scaling.