Enhancing testing cell set efficiency: A machine learning approach on hard disk drive data

Open Access

Table 5

Comparison of hyperparameters of various algorithm by default and tuned values.

Algorithm	Hyperparameter	Tuned value^a	Default value^b
LDA	n_components	0	None
	tol	0.0001	0.0001
	random_state^c	815	1
RCCV	alphas	array([1.0])	array([0.1, 1.0, 10.0])
	cv	2	None
	normalize	′deprecated′	′deprecated′
	random_state^c	416	1
ETC	criterion	′gini′	′gini′
	max_depth	11	None
	min_samples_leaf	1	1
	min_samples_split	2	2
	n_estimators	8	100
	random_state	42	None
	random_state^c	11	1
RFC	criterion	′gini′	′gini′
	max_depth	16	None
	min_samples_leaf	1	1
	min_samples_split	2	2
	n_jobs	-1	None
	n_estimators	27	100
	random_state	11	None
	random_state^c	755	1
XGBoost	max_depth	2	None
	n_estimators	45	100
	objective	′multi:softprob′	′binary:logistic′
	random_state	1	None
	random_state^c	8	1

The hyperparameters that resulted in the best accuracy scores.

The default hyperparameters in the super learner package.

The randomize parameter controls the random splitting of a dataset into training and testing sets.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.