HYPER-PARAMETERS.md

Models' hyper-parameters configuration

Decision Tree (CART)

Parameter	Possible values
criterion	'gini', 'entropy'
splitter	'best', 'random'
max_features	'auto', 'sqrt', 'log2', None
class_weight	'balanced', None

Logistic Regression

Parameter	Possible values
penalty	'l2', 'none'
tol	numpy.logspace(start=-5, stop=-3, num=10)
C	numpy.linspace(start=0, stop=2, num=10)
class_weight	'balanced', None
solver	'lbfgs', 'sag', 'saga'
fit_intercept	False, True

Naive Bayes

Parameter	Possible values
var_smoothing	numpy.logspace(start=-10, stop=-8, num=10)

Random Forest

Parameter	Possible values
n_estimators	[int(x) for x in np.linspace(start = 100, stop = 2000, num = 10)]
max_features	'auto', 'sqrt'
max_depth	[int(x) for x in np.linspace(10, 110, num = 11)], None
bootstrap	False, True

Support Vector Machine

Parameter	Possible values
C	[int(x) for x in np.linspace(start = 1, stop = 1000, num = 100)]
gamma	'scale', 'auto'
kernel	'rbf', 'sigmoid', 'poly'
degree	1, 2, 3, 4, 5
shrinking	False, True
tol	[x for x in np.logspace(start = -5, stop = -1, num = 10)]
class_weight	'balanced', None