Table 5 Number of spiked samples in 1) model training and selection, 2) cross-validation. All samples were spiked with adulterants with concentrations 0.01, 0.02, 0.05, 0.1, 0.2 g per 100 g raw milk.
From: Non-targeted detection of food adulteration using an ensemble machine-learning model
Adulterants added | Number of raw milk samples in model training and selection | Number of raw milk samples in cross-validation |
|---|---|---|
Common chemicals | ||
Potassium dichromate | 19 | 0 |
Potassium sulfate | 19 | 0 |
Sodium citrate | 4 | 0 |
Trisodium citrate | 15 | 0 |
Citric acid | 19 | 0 |
Hydrogen peroxide | 0 | 15 |
Sodium hydroxide | 0 | 15 |
Salt | 0 | 15 |
Nitrogen-based adulterants | ||
Ammonium sulfate | 15 | 0 |
Urea | 15 | 0 |
Melamine | 30 | 0 |
Skimmed milk powder | 15 | 0 |
Soy milk | 15 | 0 |
Urea | 15 | 0 |
Whey protein | 15 | 0 |
Whole milk powder | 15 | 0 |
Carbohydrate-based adulterants | ||
Sucrose | 15 | 15 |
Glucose | 15 | 15 |
Lactose | 15 | 0 |
Fructose | 15 | 15 |
Maltodextrin | 15 | 0 |
Starch | 15 | 0 |
Improperly stored for 36 h | 23 | 0 |
Water | 17 | 0 |
Cow smell | 10 | 0 |
Total | 111 | 90 |
