Bayesian Disaggregate and Aggregate Calibration of Path Logit Choice Models

Table 4

Main characteristics and utilities for the five representative users and for all samples.

(a) Attributes
Criteria	MinTime							MaxMotorway			MinCost
Utility per attribute	T, yT1,n		M, yP1,n			L, yL1,n		T, yT2,n		M, yP2,n	T, yT3,n		M, yP3,n
User 103	1.07		0.11			0		1.11		0.58	1.12		0.00
User 102	1.57		0.42			0		1.72		0.71	2.03		0.00
User 59	1.04		0.70			0		1.14		0.78	1.45		0.00
User 88	2.70		0.82			0		2.82		0.89	3.73		0.00
User 87	5.14		0.97			1		—		—	7.84		0.00
Average (186 users)	2.12		0.69			0.37		2.29		0.83	2.99		0.00
(b) Percentage contribution of the attribute component in the utilities (disaggregate, 186 users)
ID, approach	2, C							4, B
Criteria	Time			Motorway		Cost		Time			Motorway		Cost
Utility per attribute	θT⋅yT1,n	θP⋅yP1,n	θL⋅yL1,n	θT⋅yT2,n	θP⋅yP2,n	θT⋅yT3,n	θP⋅yP3,n	θT⋅yT1,n	θP⋅yP1,n	θL⋅yL1,n	θT⋅yT2,n	θP⋅yP2,n	θT⋅yT3,n	θP⋅yP3,n
User 103	71%	29%	0%	33%	67%	100%	0%	78%	22%	0%	41%	59%	100%	0%
User 102	49%	51%	0%	38%	62%	100%	0%	57%	43%	0%	47%	53%	100%	0%
User 59	27%	73%	0%	27%	73%	100%	0%	35%	65%	0%	34%	66%	100%	0%
User 88	45%	55%	0%	44%	56%	100%	0%	54%	46%	0%	53%	47%	100%	0%
User 87	13%	9%	78%	—	—	100%	0%	41%	22%	37%	66%	34%	100%	0%
Average (186 users)	13%	17%	71%	41%	59%	100%	0%	37%	33%	30%	50%	50%	100%	0%

C = Classic; B = Bayesian; T = travel time (h); M = motorway percentage ([0; 1]); L = label time (0/1); the alternative chosen is reported in bold.