Complete results and additional material for the article “CTC: Competitive in an Analysis of GBML Algorithms for Rule Induction in Imbalanced Datasets”

This page contains the complete tables which, due to space limitation, were summarized for the article “CTC: Competitive in an Analysis of GBML Algorithms for Rule Induction in Imbalanced Datasets”.

Besides, the table with the results (GM values) for the 6 resampling strategies for the CTC algorithm, Table 4, can be downloaded as an Excel document or as a CSV file.

Table 2. Description of imbalanced datasets

	#Examples	#Attributes	% Minority Class	#Examples Min.Class
Abalone19	4,174	8	0.77	32
Yeast6	1,484	8	2.49	37
Yeast5	1,484	8	2.96	44
Yeast4	1,484	8	3.43	51
Yeast2vs8	482	8	4.15	20
Glass5	214	9	4.2	10
Abalone9vs18	731	8	5.65	41
Glass4	214	9	6.07	13
Ecoli4	336	7	6.74	23
Glass2	214	9	8.78	19
Vowel0	988	13	9.01	89
Page-blocks0	5,472	10	10.23	560
Ecoli3	336	7	10.88	37
Yeast3	1,484	8	10.98	163
Glass6	214	9	13.55	29
Segment0	2,308	19	14.26	329
Ecoli2	336	7	15.48	52
New-thyroid1	215	5	16.28	35
New-thyroid2	215	5	16.89	36
Ecoli1	336	7	22.92	77
Vehicle0	846	18	23.64	200
Glass0123vs456	214	9	23.83	51
Haberman	306	3	27.42	84
Vehicle1	846	18	28.37	240
Vehicle2	846	18	28.37	240
Vehicle3	846	18	28.37	240
Yeast1	1,484	8	28.91	429
Glass0	214	9	32.71	70
Iris0	150	4	33.33	50
Pima	768	8	34.84	268
Ecoli0vs1	220	7	35	77
Wisconsin	683	9	35	239
Glass1	214	9	35.51	76
Min	150	3	0.77	10
Max	5,472	19	35.51	560
Mean	919.94	9.39	17.61	119.97
Median	482.00	8.00	15.48	52.01

Table 3. Values of number samples used for each dataset

	N_S=5	N_S 120%	N_S 200%
Abalone19	5	156	260
Yeast6	5	49	81
Yeast5	5	41	68
Yeast4	5	35	59
Yeast2vs8	5	29	49
Glass5	5	29	48
Abalone9vs18	5	22	36
Glass4	5	20	33
Ecoli4	5	18	30
Glass2	5	14	23
Vowel0	5	14	23
Page-blocks0	5	12	20
Ecoli3	5	12	19
Yeast3	5	11	19
Glass6	5	9	15
Segment0	5	9	15
Ecoli2	5	8	13
New-thyroid1	5	8	13
New-thyroid2	5	8	12
Ecoli1	5	6	9
Vehicle0	5	6	9
Glass0123vs456	5	6	9
Haberman	5	5	8
Vehicle1	5	5	8
Vehicle2	5	5	8
Vehicle3	5	5	8
Yeast1	5	5	7
Glass0	5	5	7
Iris0	5	5	7
Pima	5	5	6
Ecoli0vs1	5	5	6
Wisconsin	5	5	6
Glass1	5	5	6
Min	5	5	6
Max	5	156	260
Mean	5	17.48	28.48
Median	5	8.00	13.00

This table can be downloaded as an Excel document or as a CSV file by clicking on the following links and .

Table 4. Results (GM values) for the 6 resampling strategies for CTC algorithm

	Tam=SizeOfMinClass			Tam=Max
	N_S=5	N_S 120%	N_S 200%	N_S=5	N_S 120%	N_S 200%
Abalone19	58.99	51.03	52.63	52.38	44.44	43.43
Yeast6	83.93	83.03	83.53	82.09	81.43	79.11
Yeast5	13.42	85.29	89.95	43.31	83.53	89.40
Yeast4	79.76	79.47	77.65	77.02	75.82	74.80
Yeast2vs8	72.83	72.83	72.83	72.83	72.83	72.83
Glass5	27.60	43.56	69.75	49.84	79.64	89.69
Abalone9vs18	69.62	70.28	71.29	68.78	69.81	68.49
Glass4	16.15	43.07	42.71	47.27	76.28	67.01
Ecoli4	45.80	73.91	77.09	66.49	83.22	81.45
Glass2	57.22	60.81	60.04	65.93	64.42	62.03
Vowel0	90.46	89.36	91.89	93.83	94.33	95.14
Page-blocks0	78.82	83.19	82.92	75.03	78.74	79.38
Ecoli3	51.00	73.66	70.96	69.28	71.19	73.78
Yeast3	88.53	86.54	85.47	87.22	87.33	85.69
Glass6	89.71	88.04	87.38	87.99	86.65	85.04
Segment0	98.07	98.08	98.61	98.59	98.50	98.75
Ecoli2	83.04	84.67	85.73	87.07	85.50	87.60
New-thyroid1	89.22	94.01	93.95	95.71	95.57	95.58
New-thyroid2	93.04	93.38	93.41	94.72	94.13	94.60
Ecoli1	85.69	85.82	83.48	87.57	84.76	85.54
Vehicle0	90.84	91.11	87.46	91.98	91.72	92.24
Glass0123vs456	85.35	86.97	85.91	86.78	87.05	86.85
Haberman	49.46	47.03	47.28	50.40	49.20	50.28
Vehicle1	54.03	54.41	54.89	55.71	55.51	55.91
Vehicle2	77.71	68.79	77.95	78.21	77.33	80.96
Vehicle3	56.50	55.38	56.32	57.06	56.69	56.99
Yeast1	62.97	64.62	63.34	65.32	65.69	64.84
Glass0	75.97	79.49	79.18	80.37	78.59	79.86
Iris0	98.77	98.56	98.15	98.97	98.97	98.97
Pima	68.09	68.25	68.09	69.12	67.45	67.85
Ecoli0vs1	97.60	96.83	97.46	98.16	98.16	98.16
Wisconsin	92.02	92.52	93.69	91.82	91.93	92.17
Glass1	68.13	69.87	71.13	69.62	70.16	68.15
Mean	71.22	76.18	77.34	75.65	78.68	78.87
Avg. Rank	4.41 (6)	3.77 (5)	3.71 (4)	2.92 (2)	3.30 (3)	2.88 (1)

Table 7. Shaffer test for the comparison of the 9 algorithms

	UCS	SIA	OCEC	GASSIT	Oblique-DT	C4.5	C4.5-Rules	Ripper	CTC
UCS	x	= (1.0)	= (1.0)	= (1.0)	= (1.0)	= (1.0)	= (1.0)	- (0.0016)	= (0.4819)
SIA	= (1.0)	x	= (1.0)	= (1.0)	= (1.0)	= (1.0)	= (1.0)	- (0.0017)	= (0.5121)
OCEC	= (1.0)	= (1.0)	x	= (1.0)	= (0.8986)	= (0.5121)	= (0.2914)	- (0.0001)	= (0.0843)
GASSIT	= (1.0)	= (1.0)	= (1.0)	x	= (1.0)	= (1.0)	= (1.0)	- (0.0069)	= (0.9485)
Oblique-DT	= (1.0)	= (1.0)	= (0.8986)	= (1.0)	x	= (1.0)	= (1.0)	= (0.2914)	= (1.0)
C4.5	= (1.0)	= (1.0)	= (0.5121)	= (1.0)	= (1.0)	x	= (1.0)	= (0.5121)	= (1.0)
C4.5-Rules	= (1.0)	= (1.0)	= (0.2914)	= (1.0)	= (1.0)	= (1.0)	x	= (0.5121)	= (1.0)
Ripper	+ (0.0016)	+ (0.0017)	+ (0.0001)	+ (0.0069)	= (0.2914)	= (0.5121)	= (0.5121)	x	= (1.0)
CTC	= (0.4819)	= (0.5121)	= (0.0843)	= (0.9485)	= (1.0)	= (1.0)	= (1.0)	= (1.0)	x