| 1. |
- Bierlaire, Michel, et al.
(författare)
-
A latent route choice model in Switzerland
- 2006
-
Ingår i: Proceedings of the European Transport Conference.
-
Konferensbidrag (refereegranskat)abstract
- Rature, for example the C-Logit (Cascetta et al., 1996), Path Size Logit (Ben-Akiva and Bierlaire, 1999), Link-Nested Logit (Vovsha and Bekhor, 1998) and subnetwork (Frejinger and Bierlaire, 2006) models. A latent cho- sen route corresponds to an unobserved choice where only an approximate choice description is available. Instead of an exact route description, trav- ellers describe their choice in terms of a sequence of locations and cities that they have traversed, without the need to relate the actual network used by the analyst. We compute the probability of an (aggregate) observation with an un- derlying route choice model using a detailed network description and actual paths. In this context, not only several routes can correspond to the same observation, but the exact origin-destination pair is not necessarily known. We therefore consider several possible origin-destination pairs and their as- sociated set of routes, generated by a choice set generation algorithm. We derive from this list the probability of each observation, in order to perform the maximum likelihood estimation of the route choice model. The methodology is illustrated by estimating Path Size Logit and sub- network models using a dataset collected in Switzerland. This application is one of few based on revealed preferences (RP) data that are presented in the literature. In addition, the network used here (39411 unidirectional links and 14841 nodes) is to our knowledge the largest network used for evaluation of route choice models based on RP data. The estimation results are very sat- isfactory. Indeed, they do not only show that it is possible to estimate route choice models based on aggregate observations, but also that the parameter estimates are stable across different types of models and that the standard errors are small.
|
|
| 2. |
|
|
| 3. |
|
|
| 4. |
- Frejinger, Emma, 1979-, et al.
(författare)
-
Capturing correlation in route choice models using subnetworks
- 2006
-
Ingår i: Proceedings of the Swiss Transportation Research Conference.
-
Konferensbidrag (refereegranskat)abstract
- When using random utility models for a route choice problem, choice set generation and correlation among alternatives are two issues that make the modelling complex. In this paper we propose a modelling approach where the path overlap is captured with a subnetwork. A subnetwork is a simplification of the road network only containing easy identifiable and behaviourally relevant roads. In practise, the subnetwork can easily be defined based on the route network hierarchy. We propose a model where the subnetwork is used for defining the correlation structure of the choice model. The motivation is to explicitly capture the most important correlation without considerably increasing the model complexity. We present estimation results of a factor analytic specification of a mixture of Multinomial Logit model, where the correlation among paths is captured both by a Path Size attribute and error components. The estimation is based on a GPS dataset collected in the Swedish city of Borlänge. The results show a significant increase in model fit for the Error Component model compared to a Path Size Logit and Multinomial Logit models. Moreover, the correlation parameters are significant. We also analyse the performance of the different models regarding prediction of choice probabilities. The results show a better performance of the Error Component model compared to the Path Size Logit and Multinomial Logit models.
|
|
| 5. |
|
|
| 6. |
- Frejinger, Emma, 1979-, et al.
(författare)
-
Capturing correlation with subnetworks in route choice models
- 2007
-
Ingår i: Transportation Research Part B. - : Elsevier BV. - 0191-2615 .- 1879-2367. ; 41:3, s. 363-378
-
Tidskriftsartikel (refereegranskat)abstract
- When using random utility models for a route choice problem, a critical issue is the significant correlation among alternatives. There are basically two types of models proposed in the literature to address it: (i) a deterministic correction of the path utilities in a Multinomial Logit model (Such as the Path Size Logit or the C-Logit models) and (ii) an explicit modeling of the correlation through assumptions about the error terms, and the use of advanced discrete choice models such as the Cross-Nested Logit or the Error Component models. The first is simple, easy to handle and often used in practice. Unfortunately, it does not correctly capture the correlation structure, as we discuss in details in the paper. The second is more consistent with the modeling objectives, but very complicated to specify and estimate. The modeling framework proposed in this paper allows the analyst to control the trade-off between the simplicity of the model and the level of realism. Within this framework, the key concept capturing the correlation structure is called subnetwork. A subnetwork is a simplification of the road network only containing easy identifiable and behaviorally relevant roads. In practice, the subnetwork can easily be defined based oil the route network hierarchy. The importance and the originality of our approach lie in the possibility to capture the most important correlation without considerably increasing the model complexity. This makes it suitable for a wide spectrum of application,.;, namely involving realistic large-scale networks. As an illustration, we present estimation results of a factor analytic specification of a mixture of Multinomial Logit model, where the correlation among paths is captured by error components. The estimation is based on a GPS dataset collected in the Swedish city of Borlange. The results show a significant increase in model fit and forecasting performance for the Error Component model compared to a Path Size Logit model. Moreover, the correlation parameters are significant.
|
|
| 7. |
|
|
| 8. |
- Frejinger, Emma, 1979-, et al.
(författare)
-
Random Sampling of Alternatives for Route Choice Modeling
- 2007
-
Ingår i: Swiss Transport Research Conference.
-
Konferensbidrag (refereegranskat)abstract
- In this paper we present a new point of view on choice set generation for route choice models. When modeling route choice behavior using random utility models choice sets of paths need to be defined. Existing approaches generate paths and assume that actual choice sets are found. On the contrary, we assume that actual choice sets are the sets of all paths connecting each origindestination pair. These sets are however unknown and we propose a stochastic path generation algorithm that corresponds to an importance sampling approach. The path utilities should then be corrected according to the used sampling protocol in order to obtain unbiased parameter estimates. We derive such a sampling correction for the proposed algorithm. We present numerical results based on synthetic data. The results show that the model including sampling correction yields unbiased coefficient estimates but we also make important observations concerning the Path Size attribute. Namely, it biases the estimation results if it is not computed based on the true correlation structure. These results suggest that the Path Size attribute should be computed based on as many alternatives as possible, more than in the generated choice sets.
|
|
| 9. |
- Frejinger, Emma, 1979-, et al.
(författare)
-
Route choice modeling with network-free data
- 2008
-
Ingår i: Transportation Research Part C. - : Elsevier BV. - 0968-090X .- 1879-2359. ; 16:2, s. 187-198
-
Tidskriftsartikel (refereegranskat)abstract
- Route Choice models arc difficult to design and to estimate for various reasons. In this paper we focus on issues related to data. Indeed, real data in its original format are not related to the network used by the modeler and do therefore not correspond to path definitions. Typical examples arc data collected with the Global Positioning System (GPS) or respondents describing chosen itineraries to interviewers. Data manipulation is then necessary in order to obtain network compliant paths. We argue that such manipulations introduce bias and errors and should be avoided. We propose a general modeling framework that reconcile network-free data with a network based model without data manipulations. The concept that bridges the gap between the data and the model is called Domain of Data Relevance and corresponds to a physical area in the network where a given piece of data is relevant. We illustrate the framework on simple examples for two different types of data (GPS data and reported trips). Moreover, we present estimation results of Path Size Logit and Subnetwork models based on a dataset of reported trips collected in Switzerland. The network is to our knowledge the largest one used in the literature for route choice analysis based on revealed preferences data.
|
|
| 10. |
|
|
| 11. |
|
|
| 12. |
|
|
