Software Development Effort Estimation Using Regression Fuzzy Models

Table 10

Error measures and meaningfulness tests for datasets without outliers.

	MAE	MBRE	MIBRE	SA	Δ	ME
	Dataset 1
MLR_out	1518.4	72.4	241.7	36.1	0.3	−296.5
Fuzzy Lin_out	720	26.5	39.3	69.7	0.6	26.6
Fuzzy Const_out	1111.3	255.6	44.8	53.2	0.4	−214.5
Fuzzy Mam_out	2834	330.1	56.6	−19.2	0.2	−2774.5
	Dataset 2
MLR_out	1418.6	26.1	19.2	80.9	0.9	−910.2
Fuzzy Lin_out	1342.9	21	16.3	81.9	0.9	−801.6
Fuzzy Const_out	3674.7	85.8	40.2	50.5	0.5	2268.4
Fuzzy Mam_out	3268.8	92.8	37.1	56	0.6	−2219
	Dataset 3
MLR_out	4742.1	−2.2	33.6	53.2	0.5	513.4
Fuzzy Lin_out	4376.3	−114.9	31.9	56.8	0.6	−528.6
Fuzzy Const_out	4187.5	66.7	28.7	58.7	0.6	2891.3
Fuzzy Mam_out	5608.5	70.7	35.8	44.7	0.5	−1523.9
	Dataset 4
MLR_out	3982	333.7	50	32.2	0.3	−1673
Fuzzy Lin_out	3613.7	181.8	62.5	38.5	0.4	−1287
Fuzzy Const_out	4377.7	421.5	56.1	25.4	0.3	−1551
Fuzzy Mam_out	5897.6	348.2	55.9	−0.4	0	−3807

Note: MAE: mean absolute error; SA: for standardized; Δ (delta): effect size, MBRE: mean balance relative, MIBRE: mean inverted balance relative error.