An empirical study on aggregation of alternatives and its influence on prediction in car type choice models

Abstract

Assessing and predicting car type choices are important for policy analysis. Car type choice models are often based on aggregate alternatives. This is due to the fact that analysts typically do not observe choices at the detailed level that they are made. In this paper, we use registry data of all new car purchases in Sweden for two years where cars are observed by their brand, model and fuel type. However, the choices are made at a more detailed level. Hence, an aggregate (observed) alternative can correspond to several disaggregate (detailed) alternatives. We present an extensive empirical study analyzing estimation results, in-sample and out-of-sample fit as well as prediction performance of five model specifications. These models use different aggregation methods from the literature. We propose a specification of a two-level nested logit model that captures correlation between aggregate and disaggregate alternatives. The nest specific scale parameters are defined as parameterized exponential functions to keep the number of parameters reasonable. The results show that the in-sample and out-of-sample fit as well as the prediction performance differ. The best model accounts for the heterogeneity over disaggregate alternatives as well as the correlation between both disaggregate and aggregate alternatives. It outperforms the commonly used aggregation method of simply including a size measure.

Combined Vehicle Type and Fuel Type Choices of Private Households: An Empirical Analysis for Germany

“…We further find that respondents who place more emphasis on the motor vehicle tax during new car purchase decisions are less likely to buy vehicles that run on diesel, which is in line with Habibi et al (2019). Our result makes sense, as in Germany this tax for diesel vehicles is higher, compared to gasoline vehicles and AFVs.…”

“…Hence, several studies have jointly considered vehicle type and fuel type choices in their explanatory models in order to account for this heterogeneity in vehicle buyers' preferences (e.g. Habibi et al, 2019;Hahn et al, 2018;Liu and Cirillo, 2018;DeShazo et al, 2017;Higgins et al, 2017;Dumortier et al, 2015;Mabit, 2014;Hess et al, 2012;Potoglou and Kanaroglou, 2007).…”

The evolution of choice set formation in dwelling and location with rising prices: A decadal panel analysis in the Greater Toronto Area

“…However, the need to estimate 2 K possible choice sets introduces computational challenges as it requires large matrices to be stored in memory. It is proposed to use an aggregated choice set and the method used by Habibi et al (2019) for car type choice. An aggregate choice set is defined as comprising aggregated locations and dwelling types, with disaggregate choice alternatives represented through a set of statistics.…”

Modelling long-distance route choice using mobile phone call detail record data: a case study of Senegal

“…unique route choices for some travellers while broad sub-groups of the possible routes for the rest). Although there is no specific route choice model dealing with such a scenario, recent advances in other fields of choice modelling have addressed this challenge using the broad choice framework (proposed by Wong, 2015 and later used by Habibi et al 2017, Lloro and Brownstone 2018, Wong et al 2018, Yip et al2018 in the context of vehicle choice and by Tran et al 2019 in the context of hotel choice), which can be adapted to deal with the current route choice modelling scenario, discussed later in this paper. It may be noted that according to the choice modelling literature, this framework is the only one that can deal with the case where the choices are uniquely observed only for some respondents and for the rest, only the sub-group of choices are observed.…”