Incremental Optimization Mechanism for Constructing a Decision Tree in Data Stream Mining

Table 1

(a) Comparison of VFDT using different 𝜏  and 𝑛_min, (b) comparison of VFDT using different 𝑛_min.

(a)

𝜏	0.10	0.20	0.30	0.40	0.50	0.60	0.70	0.80	0.90	1.00
LED24 (𝑛min=200)
Accuracy (%)	75.88	76.97	77.14	77.42	77.47	77.50	77.56	77.56	77.56	77.56
No. leaf	143.00	522.00	1124.00	1857.00	2618.00	3723.00	3743.00	3743.00	3743.00	3743.00
Time (sec)	8.70	9.91	10.92	11.51	11.92	12.78	12.32	12.32	12.43	12.45
Waveform21 (𝑛min=200)
Accuracy	85.00	86.53	86.61	86.72	86.72	86.72	86.72	86.72	86.72	86.72
No. leaf	506.00	1565.00	2492.00	2623.00	2623.00	2623.00	2623.00	2623.00	2623.00	2623.00
Time	17.80	18.50	18.89	18.69	18.77	18.72	18.53	18.72	18.74	18.72

(b)

𝑛min	200	300	400	500	600	700	800	900	1000
LED24 (𝜏=0.7)
Accuracy	77.5611	77.5867	77.4565	77.3472	77.2557	77.1417	77.1412	77.0847	76.9887
No. leaf	3743	2405	1826	1383	1244	1057	935	804	689
Time	11.66887	10.93567	10.40527	10.07766	9.469261	9.42246	9.032458	9.172859	8.642455
Waveform21 (𝜏=0.4)
Accuracy	86.7218	86.5226	86.3028	85.9499	85.9119	85.6378	85.6707	85.7318	85.2165
No. leaf	2623	1800	1363	1103	940	806	703	644	572
Time	18.31452	17.70611	17.34731	17.03531	16.84811	16.58291	16.61411	16.61411	16.2709