Table 6.
The classification performance of CART, LDA and PLSDA on data fused with two different strategies.
| Methods | Real | Predicted | Error rate (%) | Accuracy rate (%) | |||
|---|---|---|---|---|---|---|---|
| CP | LT | Not assigned | |||||
| CART | Augment | CP | 72 | 1 | 0 | 37 | 86 |
| LT | 11 | 4 | 0 | ||||
| PCA | CP | 68 | 5 | 0 | 50 | 78 | |
| LT | 14 | 1 | 0 | ||||
| LDA | Augment | CP | 64 | 9 | 0 | 29 | 82 |
| LT | 7 | 8 | 0 | ||||
| PCA | CP | 69 | 4 | 0 | 49 | 80 | |
| LT | 14 | 1 | 0 | ||||
| PLSDA | Augment(12) | CP | 62 | 11 | 0 | 14 | 85 |
| LT | 2 | 13 | 0 | ||||
| PCA(1) | CP | 46 | 27 | 0 | 42 | 61 | |
| LT | 7 | 8 | 0 | ||||
CP is short for Codonopsis pilosula (Franch.) Nannf; LT is short for Codonopsis pilosula Nannf. var. modesta (Nannf. ) L. T. Shen; the number in the parenthesis denotes the number of latent variables optimized by eight-fold cross-validation.