Cross-Nested Logit Model for the Joint Choice of Residential Location, Travel Mode, and Departure Time

doi:10.13209/j.0479-8023.2017.066

Abstract

Abstract:

This paper aims to describe the joint choice of trip chaining, travel mode, and departure time. First, based on random utility maximization theory, the innovative Cross-Nested Logit model and traditional NL model are formulated respectively. Travel time, travel cost, and factors depicting the individual and family socio-economic characteristics are defined as exogenous variables, and the model choice sets are the combination of trip chaining subset, departure time subset, and travel mode choice subset. Second, using Beijing traffic survey data of 2010, the model parameters are estimated, and the direct and cross elasticity are calculated to analyze the change of alternatives probability brought by factors variation. Estimation results show the Cross-Nested Logit model is more accurate statistically than any kind of NL model. Estimation results show that decision makers will change their departure time in the first place, followed by mode choice, and finally, trip chaining type, when exogenous variables alter. Moreover, elasticity analysis results reveal that car travelers of complex trip chaining are less sensitive to travel time and travel cost than car traveler of simple trip chaining.

Key words: trip chaining, travel mode, departure time, joint choice, cross-nested logit, GEV

摘要：

为刻画出行链、出行方式与出发时间的联合选择行为, 选取出行耗时、出行费用及个人与家庭属性等作为效用变量, 以出行链选择子集合、出行方式选择子集合和出发时间选择子集合的组合作为模型的选择项, 构建基于广义极值(GEV)理论的交叉巢式Logit模型, 为方便对比, 同时构建3种结构的传统NL模型。利用2010年北京市居民工作出行的小样本调查数据, 对模型参数进行估计和检验, 并进行弹性分析, 分析效用变量的改变对备选方案选择概率的影响。参数估计结果表明, 交叉巢式Logit模型具有比NL模型更优的统计学特征, 并发现当效用变量改变时, 选择者最先变更其出发时间, 然后是出行方式, 最后才考虑改变其出行链结构。直接和交叉弹性分析表明, 与简单链的小汽车出行者相比, 复杂链的小汽车出行者对出行时间与出行费用的敏感性较低。研究结果可以为制定和评价交通需求管理政策提供依据。

关键词: 出行链, 出行方式, 出发时间, 联合选择, 交叉巢式Logit, 广义极值(GEV)理论

CLC Number:

U491

Liya YANG, Juan LI. Cross-Nested Logit Model for the Joint Choice of Residential Location, Travel Mode, and Departure Time[J]. Acta Scientiarum Naturalium Universitatis Pekinensis, 2017, 53(4): 722-730.

杨励雅, 李娟. 居民出行链、出行方式与出发时间联合选择的交叉巢式Logit模型[J]. 北京大学学报自然科学版, 2017, 53(4): 722-730.

Figures/Tables 6

References 24

[1]	Shiftan Y.Practical approach to model trip chaining. Transportation Research Record 1645. Journal of Transportation Research Board, National Research Council, Washington DC, 1998, 1645(1): 17-23
[2]	Dissanayake D, Morikawa T.Household travel behavior in developing countries: nested logit model of vehicle ownership, mode choice, and trip chaining. Transportation Research Record 1645. Journal of Transportation Research Board, National Research Council, Washington DC, 2002, 1805(1): 45-52
[3]	Strathman J G, Dueker K J. Understanding trip chaining. Special reports on trip and vehicle attributes[R].1990 NPTS Reports Series. Publication No.FHWA- PL-95-033. US Department of Transportation, 1995: 1-1-1-27
[4]	Ye X, Pendyala R M, Gottardi G.An exploration of the relationship between mode choice and complexity of trip chaining patterns. Transportation Research Part B, 2007, 41: 96-113
[5]	杨敏, 陈学武, 王炜, 等. 通勤出行简单链和复杂链的选择行为研究. 武汉理工大学学报: 交通科学与工程版, 2008, 32(2): 191-194
[6]	Bhat C R.Worker travel mode choice and number of non-work commute stops. Transportation Research Part B, 1997, 31(1): 41-54
[7]	Strathman J G, Dueker K J.Effect of household structure and selected characteristics on trip chaining. Transportation, 1994, 21: 23-45
[8]	Albert A.Mode choice for the journey to work among formal sector employees in Accra, Ghana. Journal of Transport Geography, 1993, 1(4): 219-229
[9]	Wafaa S.Implications of congestion charging for departure time choice: work and non-work schedule flexibility. Transportation Research Part A, 2005, 39: 773-791
[10]	Koppelman F S, Sethi V.Closed-form discrete choice models // Hensher D A, Button K J. Handbook of transport Modeling, 2nd ed. Amsterdam: Elsevier Science, 2008: 211-225
[11]	Hensher D A, Reyes A J.Trip chaining as a barrier to the propensity to use public transport. Transportation, 2000, 27(4): 341-361
[12]	Bekhor S, Prashker J N.GEV-based destination choice models that account for unobserved similarities among alternatives. Transportation Research Part B, 2008, 42(3): 243-262
[13]	Kenneth A.Small. Approximate generalized extreme value models of discrete choice. Journal of Econometrics, 1994, 62(2): 351-382
[14]	Swait J.Choice set generation within the generalized extreme value family of discrete choice models. Trans- portation Research Part B, 2001, 35(7): 643-666
[15]	Bhat C R.Analysis of travel mode and departure time choice for urban shopping trips. Transportation Research Part B, 1998, 32(6): 361-371
[16]	Wen C H, Koppelman F S.The generalized nested logit model. Transportation Research Part B, 2001, 35: 627-641
[17]	Vovsha P.The cross-nested logit model: application to mode choice in the Tel-Aviv metropolitan area. Transportation Research Record, 1997, 1607: 6-15
[18]	Bierlaire M.A theoretical analysis of the cross-nested logit model. Ann Operation Research, 2006, 144: 287-300
[19]	Koppelman F S, Wen C H.The paired combinatorial logit model: properties, estimation and application. Transportation Research Part B, 2000, 34(2): 75-89
[20]	Bhat C R, Guo J Y.A mixed spatially correlated logit model: formulation and application to residential choice modeling. Transportation Research Part B, 2004, 38(2): 147-168
[21]	Hess S, Polak J W.Exploring the potential for cross-nesting structures in airport-choice analysis: a case-study of the Greater London area. Transportation Research Part E, 2006, 42: 63-81
[22]	Hess S, Daly A, Rohr C, et al.On the development of time period and mode choice models for use in large scale modeling forecasting systems. Transportation Research Part A, 2007, 41: 802-826
[23]	Bierlaire M.BIOGEME: a free package for the estimation of discrete choice models // The Third Swiss Transportation Research Conference. Ascona, 2003: 1-24
[24]	Papola A.Some developments on the cross-nested logit model. Transportation Research Part B, 2004, 38: 833-851

变量名		说明
出行总时间	TT	连续变量, 一次出行链耗费的时间(min)
出行总费用	TC	连续变量, 一次出行链的总费用, 小汽车出行费用以与距离相关的油费表示, 公共交通则以票价表示
往返行程距离	Dis	出行链各部分距离之和
年龄	Age1	哑变量, 1为年龄25岁及以下
	Age2	哑变量, 1为年龄在26~55之间,
	Age3	哑变量, 1为年龄在56及以上
收入	Income1	哑变量, 1为月收入低于5000 RMB
	Income2	哑变量, 1为月收入在5001~10000 RMB之间
	Income3	哑变量, 1为月收入大于10000 RMB
是否拥有驾照	License	哑变量, 1为是, 0为否
工作时间是否弹性	Time	哑变量, 1为是, 0为否
出行者在家活动时间(睡眠除外)	Duin	连续变量, 单位为小时
家中是否有12岁以下儿童	Kids	哑变量, 1为是, 0为否
家庭拥有车辆数	Car	离散变量, 家庭拥有小汽车台数

变量名		说明
出行总时间	TT	连续变量, 一次出行链耗费的时间(min)
出行总费用	TC	连续变量, 一次出行链的总费用, 小汽车出行费用以与距离相关的油费表示, 公共交通则以票价表示
往返行程距离	Dis	出行链各部分距离之和
年龄	Age1	哑变量, 1为年龄25岁及以下
	Age2	哑变量, 1为年龄在26~55之间,
	Age3	哑变量, 1为年龄在56及以上
收入	Income1	哑变量, 1为月收入低于5000 RMB
	Income2	哑变量, 1为月收入在5001~10000 RMB之间
	Income3	哑变量, 1为月收入大于10000 RMB
是否拥有驾照	License	哑变量, 1为是, 0为否
工作时间是否弹性	Time	哑变量, 1为是, 0为否
出行者在家活动时间(睡眠除外)	Duin	连续变量, 单位为小时
家中是否有12岁以下儿童	Kids	哑变量, 1为是, 0为否
家庭拥有车辆数	Car	离散变量, 家庭拥有小汽车台数

变量	以出发时间嵌套		以出行方式嵌套			以出行链嵌套
变量	参数值	t-stat	参数值	t-stat		参数值	t-stat
出行总时间TT	-1.927^a	-10.0	-1.461^a	-9.8		-1.084^a	-9.9
出行总费用TC	-1.166^a	-8.5	-1.030^a	-7.9		-0.946^a	-7.4
往返行程距离Dis	0.0000	0.3	-0.0001	-0.4		0.0009^a	2.2
Age1	-0.173^a	-4.1	-0.105^a	-3.1		-0.092^a	-3.1
Age3	-0.0032	-0.8	-0.0018	-0.7		-0.0004	-0.2
Income2	0.0688^a	2.3	0.0235	1.8		0.0213	1.8
Income3	-0.0001	-1.8	0.0004^a	2.3		0.0000	0.9
是否拥有驾照	0.530^a	2.9	0.552^a	2.9		0.389^a	2.4
工作时间是否弹性	-0.238^a	-3.2	-0.364^a	-3.4		-0.215^a	-3.2
出行者在家活动时间(睡眠除外)	-7.701^a	-5.8	-7.012^a	-5.5		-6.884^a	-5.0
家中是否有12岁以下儿童	0.683^a	3.2	0.716 ^a	4.0		0.717^a	4.0
家庭拥有车辆数	2.001^a	7.3	2.896^a	7.7		2.092^a	7.4
异质参数μ	早高峰		步行/自行车			简单链
	早高峰		0.98		0.0
	1.0	—	公共交通
	中间时段		0.70		1.9	0.66^b	2.9
	0.84	0.6	小汽车			复杂链
	晚高峰		0.62^b		2.5
	晚高峰		组合方式
	0.89	0.2	0.79		1.1	0.58^b	3.7
调整 ρ²	0.319		0.334			0.372
样本数	2244		2244			2244

变量	以出发时间嵌套		以出行方式嵌套			以出行链嵌套
变量	参数值	t-stat	参数值	t-stat		参数值	t-stat
出行总时间TT	-1.927^a	-10.0	-1.461^a	-9.8		-1.084^a	-9.9
出行总费用TC	-1.166^a	-8.5	-1.030^a	-7.9		-0.946^a	-7.4
往返行程距离Dis	0.0000	0.3	-0.0001	-0.4		0.0009^a	2.2
Age1	-0.173^a	-4.1	-0.105^a	-3.1		-0.092^a	-3.1
Age3	-0.0032	-0.8	-0.0018	-0.7		-0.0004	-0.2
Income2	0.0688^a	2.3	0.0235	1.8		0.0213	1.8
Income3	-0.0001	-1.8	0.0004^a	2.3		0.0000	0.9
是否拥有驾照	0.530^a	2.9	0.552^a	2.9		0.389^a	2.4
工作时间是否弹性	-0.238^a	-3.2	-0.364^a	-3.4		-0.215^a	-3.2
出行者在家活动时间(睡眠除外)	-7.701^a	-5.8	-7.012^a	-5.5		-6.884^a	-5.0
家中是否有12岁以下儿童	0.683^a	3.2	0.716 ^a	4.0		0.717^a	4.0
家庭拥有车辆数	2.001^a	7.3	2.896^a	7.7		2.092^a	7.4
异质参数μ	早高峰		步行/自行车			简单链
	早高峰		0.98		0.0
	1.0	—	公共交通
	中间时段		0.70		1.9	0.66^b	2.9
	0.84	0.6	小汽车			复杂链
	晚高峰		0.62^b		2.5
	晚高峰		组合方式
	0.89	0.2	0.79		1.1	0.58^b	3.7
调整 ρ²	0.319		0.334			0.372
样本数	2244		2244			2244

变量		参数值	t-stat
出行总时间TT		-0.949^a	-11.1
出行总费用TC		-0.825^a	-8.1
往返行程距离Dis		0.0003	0.6
Age1		-0.035^a	-3.6
Age3		-0.0002	-0.2
Income2		0.0513^a	3.4
Income3		0.0084	1.8
是否拥有驾照		0.582^a	4.2
工作时间是否弹性		-1.964	-4.4
出行者在家活动时间(睡眠除外)		-8.205	-7.3
家中是否有12岁以下儿童		0.691	5.2
家庭拥有车辆数		1.992	8.0
调整ρ²		0.401
样本数		2244
异质参数μ	早高峰	0.62^b	2.2
	中间时段	0.73^b	1.6
	晚高峰	0.59	2.4
	步行/自行车	0.34^b	3.3
	公共交通	0.45^b	2.9
	小汽车	0.29^b	3.9
	组合方式	0.57	1.8
	简单链	0.16^b	9.7
	复杂链	0.12^b	14.8