Optimization of a Semiflexible Demand-Responsive Feeder System in Suburban Areas Using a Memetic Algorithm

Table 4

Sensitivity analysis of the MA for a population of 50, 100, or 200 solutions and the stop condition without improvements of the best solution for 20, 50, and 100 iterations. Results show the passenger travel time (t.t.) and the computation time (CPU).

Max. number of iterations without improvements
Population sizes		20		50		100
		t.t. (min)	CPU (s)	t.t. (min)	CPU (s)	t.t. (min)	CPU (s)
50	30%	33.5	18	33.8	34	33.2	69
	50%	31.7	16	31.7	38	31.1	64
	70%	31.3	17	31.4	34	31.2	63
	Avg	32.2	17	32.3	36	31.8	66
100	30%	33.7	29	33.4	60	33.1	128
	50%	31.4	29	31.1	54	31.3	125
	70%	31.1	29	30.9	52	31	143
	Avg	32.1	29	31.8	55	31.8	132
200	30%	33.4	57	33.4	122	32.8	249
	50%	31	58	31.3	118	30.8	193
	70%	30.4	62	30.7	138	30.7	242
	Avg	31.6	59	31.8	126	31.4	228