Benchmarking in Data Envelopment Analysis: An Approach Based on Genetic Algorithms and Parallel Programming

Table 1

Execution cost and % of valid chromosomes when the four methods of initiation are used, all when varying the problem size.

Size			Method 1		Method 2		Method 3		Method 4
𝑚	𝑛	𝑠	Time	% val.	Time	% val.	Time	% val.	Time	% val.
2	15	1	0.0030.004	0.751.71	0.0080.005	50.8341.92	26.42351.440	82.0838.58	17.2443.566	10.583.12
3	25	2	0.0040.004	0.000.00	0.0100.006	33.5538.24	6.72216.025	90.0530.46	21.2830.801	0.800.83
4	30	2	0.0040.005	0.000.00	0.0220.008	26.8729.09	0.2230.584	100.000.00	29.52110.540	1.571.20
5	40	3	0.0040.003	0.000.00	0.0190.003	13.9023.90	13.12520.640	73.9043.40	18.1871.209	0.000.00
6	60	4	0.0060.000	0.000.00	0.0320.001	0.030.16	2.0661.132	34.7444.07	103.6980.185	0.180.46
10	100	10	0.0110.000	0.000.00	0.0920.002	0.000.00	8.4263.235	32.6541.44	302.3160.713	0.000.00