Research Article
Feasibility Study on the Use of a Portable NIR Spectrometer and Multivariate Data Analysis to Discriminate and Quantify Adulteration in Fertilizer
Table 2
The influence of preprocessing technique on identification models (fertilizer A).
| Models | Evaluation | Preprocessing treatment at 5PCs | Raw | SNV | FD | SD | Train | Test | Train | Test | Train | Test | Train | Test |
| RF | Accuracy | 93.70 | 94.70 | 93.50 | 94.60 | 96.50 | 96.50 | 93.70 | 94.70 | Error rate | 6.30 | 5.30 | 6.50 | 5.40 | 3.50 | 3.50 | 6.30 | 5.30 | Sensitivity | 96.40 | 96.00 | 95.30 | 94.00 | 96.80 | 97.60 | 96.40 | 96.00 | Specificity | 99.70 | 99.70 | 98.90 | 99.50 | 98.50 | 99.20 | 99.70 | 99.70 |
| SVM | Accuracy | 92.30 | 92.30 | 98.50 | 98.50 | 99.20 | 99.50 | 92.30 | 92.30 | Error rate | 7.70 | 7.70 | 1.50 | 1.50 | 0.80 | 0.50 | 7.70 | 7.70 | Sensitivity | 96.30 | 97.10 | 98.50 | 96.80 | 99.20 | 99.50 | 96.30 | 97.10 | Specificity | 92.30 | 92.30 | 97.50 | 97.50 | 99.20 | 99.50 | 92.30 | 92.30 |
|
|