Table 2.

The result of ANFIS based classification models.

Measure	Formula	ANFIS Algorithms
		GA-ANFIS				PSO-ANFIS				GD-ANFIS
		Pearson based		PCA based		Pearson based		PCA based		Pearson based		PCA based
		Train	Test	Train	Test	Train	Test	Train	Test	Train	Test	Train	Test
Accuracy (Acc)		0.932	0.917	0.926	0.92	0.985	0.986	0.961	0.954	0.997	0.98	0.98	0.968
Precision		0.934	0.920	0.925	0.920	0.985	0.986	0.963	0.957	0.998	0.994	0.979	0.968
Recall		0.932	0.917	0.926	0.92	0.985	0.986	0.961	0.954	0.997	0.988	0.979	0.972
F1 Score		0.932	0.917	0.926	0.919	0.985	0.983	0.963	0.954	0.996	0.988	0.979	0.968
Specificity		0.996	0.958	0.962	0.960	0.992	0.993	0.980	0.977	0.998	0.994	0.989	0.984
Error rate		0.044	0.054	0.048	0.053	0.009	0.008	0.022	0.03	0.0018	0.006	0.013	0.02
Type 1 error		0.03	0.04	0.036	0.039	0.006	0.006	0.021	0.022	0.0042	0.005	0.009	0.0153
Type 2 error		0.067	0.082	0.073	0.071	0.015	0.013	0.037	0.045	0.008	0.011	0.019	0.029

In this study, correctly identified mastitis instances in the prediction model were categorized as “True Positives” (TP). Instances incorrectly classified as mastitis, when in fact they were not, were designated as “False Positives” (FP). Undetected mastitis instances were labeled as “False Negatives” (FN), and correctly identified non-mastitis instances were classified as “True Negatives” (TN)

Significant values are in [bold].