Adaptive Ensemble with Human Memorizing Characteristics for Data Stream Mining

Table 1

Data sets description.

Dataset	Number of insts.	Number of attrs.	Number of cls.	Noise	Number of drifts	Drift type	Application area
SEA	1,000,000	3	4	10%	3	Sudden	Artificial data
Tree	100,000	10	6	0%	15	Recurring	Artificial data
Hyp	1,000,000	10	2	5%	1	Gradual	Artificial data
RBF	1,000,000	20	4	0%	4	Gradual	Artificial data
LED	1,000,000	24	10	10%	3	Gradual	Artificial data
Adult	48,842	14	2			Unknown	Census Income
Bank	41,188	20	2			Unknown	Bank marketing
Conn	67,557	42	3			Unknown	Connect-4
Cover	581,012	53	7			Unknown	Covertype
Elec	45,312	7	2			Unknown	Electricity market
EEG	14,980	14	2			Unknown	EEG Eye State
page	5,473	10	5			Unknown	Page blocks
Person	164,860	7	11			Unknown	Person Activity
Poker	1,025,010	10	10			Unknown	Poker game
Robot	5,456	24	4			Unknown	Robot