Table 6.
Summary of the partial least squares discriminant analysis (PLS-DA) and partial least squares support vector machines (PLS-SVM) analysis. The described datasets were also used in principal component analysis (PCA).
| Analysis | Treatments | Dataset Size | PLS-DA | SVM | Accuracy (%) | Confusion Matrix | |||
|---|---|---|---|---|---|---|---|---|---|
| Var (%) | RMSECV | c | Gamma | Ts | CV | ||||
| Pot experiment: | |||||||||
| Treatment—pooled | BfL:BfH:PC:NC | 119 | 32.50 | 0.356 | 0.34 | 0.03 | 96.3 | 87.2 | Table S6 |
| Untreated vs. treated | NC:PC, BfL, BfH | 119 | 82.16 | 0.192 | 0.01 | 0.01 | 100 | 100 | Table S7 |
| B. firmus-inoculated vs. non-inoculated | BfL, BfH:NC, PC | 119 | 56.50 | 0.319 | 0.01 | 0.02 | 100 | 97.4 | Table S8 |
| B. firmus inoculum size | BfL:BfH | 64 | 55.44 | 0.274 | 0.01 | 0.01 | 100 | 100 | Table S9 |
| Microplot experiment: | |||||||||
| Treatment—pooled | BfH:PC:NC | 82 | 79.10 | 0.298 | 0.24 | 0.02 | 98.9 | 96.3 | Table S10 |
Var—explained variance of the selected PLS components; RMSECV—root mean squared error of cross-validation of selected PLS components; c—SVM cost of classification parameter; gamma—SVM Gaussian kernel parameter; Ts—training set; CV—cross-validation; NC—negative control; PC—positive control (nematicide); BfL—low B. firmus inoculum; BfH—high B. firmus inoculum.