chemmodlab: a cheminformatics modeling laboratory R package for fitting and assessing machine learning models

Table 1 The best performing D–M combinations according to all performance measures considered

Descriptor	Model	IE 300	IE 100	Spec	Sens	Error Rate	PPV	F1	AUC
Burd	RF	\(\checkmark\)	\(\checkmark\)	\(\checkmark\)	\(\checkmark\)	\(\checkmark\)	\(\checkmark\)	\(\checkmark\)	\(\checkmark\)
Burd	SVM	\(\checkmark\)	\(\checkmark\)	\(\checkmark\)	\(\checkmark\)	\(\checkmark\)		\(\checkmark\)
Phar	RF	\(\checkmark\) –	\(\checkmark\)	\(\checkmark\)		\(\checkmark\)	\(\checkmark\)	\(\checkmark\) –	\(\checkmark\) –
Burd	KNN	\(\checkmark\)	\(\checkmark\)	\(\checkmark\)					\(\checkmark\)
Phar	SVM			\(\checkmark\)		\(\checkmark\)	\(\checkmark\)

Check indicates the D–M combination was among the best performers according to a performance measure using a significance level of 0.05. Check minus indicates marginal significant difference between the D–M combination and the best performer (significance level between .01 and .05). Performance measures considered were: initial enhancement at 300 tests, initial enhancement at 100 tests, specificity, sensitivity, error rate, positive predictive value, F1 measure, and area under the receiver operating characteristic curve

ISSN: 1758-2946