Table 3 Accuracy results of four different supervised learning models with five windows.

From: Rapid analysis of meat floss origin using a supervised machine learning-based electronic nose towards food authentication

Model

Accuracy (%)

Maximum

Minimum

Mean

Median

Validation

Testing

Validation

Testing

Validation

Testing

Validation

Testing

LDA

99.9

100

99.7

100

99.6

100

99.4

100

QDA

97.4

93.0

95.9

99.0

98.5

93.0

96.0

92.0

k-NN

94.8

92.0

97.0

93.3

93.4

89.3

93.3

92.0

RF

98.6

97.3

98.7

100

97.5

96.0

97.5

100

  1. The four models were analyzed by different characteristic extraction methods (maximum, minimum, mean, and median). Validation is generated with 75% of the overall data to train the model, while testing is generated to evaluate the model using 25% of the overall data.
Back to article page