Application of sample balance-based multi-perspective feature ensemble learning for prediction of user purchasing behaviors on mobile wireless network platforms

EURASIP Journal on Wireless Communications and Networking

Table 1 Main parameter settings of the models

Model type	Main parameter		Parameter value
XL	booster		gbtree
	n_estimator		500
	learning_rate		0.05
	gamma		0
	subsample		0.5
	colsample_bytree		0.8
	max_depth		10
	eval_metric		logloss
	min_child_weight		6
LL	boosting_type		gdbt
	learning_rate		0.1
	num_leave		50
	max_depth		6
	num_leaves		64
	min_child_samples		20
	min_child_weight		0.002
	reg_lambda		0.03
CDC	type of forests		Completely random tree forest, random forest
	Multi-grained scanning	Forests	2
		Trees in each forest	1000
		Sliding window size	{⌊151/16⌋, ⌊151/18⌋, ⌊151/4⌋}
	Cascade	Forests	8
	Cascade	Trees in each forest	500