QSAR-Co-X: an open source toolkit for multitarget QSAR modelling

Table 2 Hyper-parameters tuning options available in QSAR-Co-X toolkit

Technique	Parameters tuning^a
RF	Bootstrap: True/ False^b
	Criterion: Gini, Entropy,
	Maximum depth: 10, 30, 50, 70, 90, 100, 200, None
	Maximum features: Auto, Sqrt
	Minimum samples leaf: 1, 2, 4
	Minimum samples split: 2, 5, 10
	Number of estimators: 50, 100, 200,500
kNN	Number of neighbours: 1–50
	Weight options: Uniform, Distance
	Algorithms: Auto, Ball tree, kd_tree, brute
Bernoulli NB	Alpha:1, 0.5, 0.1
Bernoulli NB	Fit_prior: True, False
SVC	C: 0.1, 1, 10, 100, 1000
	Gamma: 1, 0.1, 0.01, 0.001
	Kernel: RBF, Linear, Poly, Sigmoid
MLP	Hidden layer sizes: To be specified by the user
	Activation: Identity, Logistic, Tanh, Relu
	Solver: SGD, Adam
	Alpha: 0.0001, 0.001, 0.01, 1
	Learning rate: Constant, Adaptive, Invscaling
GB	Loss: deviance, exponential
	Learning rate: 0.01, 0.05, 0.1, 0.2
	Min samples split: 0.1,0.2,0.3,0.4,0.5
	Minimum samples leaf: 0.1,0.2,0.3,0.4,0.5
	Maximum depth: 3,5,8
	Maximum features: Log2, Sqrt
	Criterion: Friedman MSE, MAE
	Subsample: 0.5, 0.6, 0.8
	Number of estimators: 50,100,200,300

^aFor further details on these parameters, check the manual associated with the toolkit in https://github.com/ncordeirfcup/QSAR-Co-X
^bThis option is automatically selected

ISSN: 1758-2946