Table 5.

Decision matrix of alternatives for the BIRADS dataset.

Criteria	Accuracy	Recall	Precision	F1-Score	ROC AUC	Log Loss	Number of Training Samples Needed	Impact of Feature Scaling	Impact of Hyperparameter Tuning	Tolerance to İrrelevant Attributes
SVM	97.0%	95.5%	97.5%	98.5%	99.5%	−0.8110	0.92	0.92	YES	0.92
Random Forest	96.0%	96.0%	98.0%	98.0%	99.0%	−0.8026	0.75	0.08	YES	0.08
Logistic Regression	95.5%	95.5%	97.0%	96.5%	99.0%	−0.7984	0.50	0.25	NO	0.50
KNN	95.5%	96.0%	97.5%	96.0%	98.5%	−0.7990	0.08	0.92	YES	0.50
Naive Bayes	94.0%	94.0%	96.0%	96.0%	98.0%	−0.7860	0.50	0.08	NO	0.75