A new logit-artificial neural network ensemble for mode choice modeling: a case study for border transport Uneb Gazder* and Nedal T. Ratrout Department of Civil and Environmental Engineering, King Fahd University of Petroleum and Minerals, Dhahran 31261, Saudi Arabia SUMMARY Logit model is one of the statistical techniques commonly used for mode choice modeling, while artificial neural network (ANN) is a very popular type of artificial intelligence technique used for mode choice modeling. Ensemble learning has evolved to be very effective approach to enhance the performance for many applications through integration of different models. In spite of this advantage, the use of ANN- based ensembles in mode choice modeling is under explored. The focus of this study is to investigate the use of aforementioned techniques for different number of transportation modes and predictor variables. This study proposes a logit-ANN ensemble for mode choice modeling and investigates its efficiency in different situations. Travel between Khobar-Dammam metropolitan area of Saudi Arabia and Kingdom of Bahrain is selected for mode choice modeling. The travel on this route can be performed mainly by air travel or private vehicle through King Fahd causeway. The results show that the proposed ensemble gives consistently better accuracies than single models for multinomial choice problems irrespective of number of input variables. Copyright © 2015 John Wiley & Sons, Ltd. KEY WORDS: mode choice modeling; logit model; artificial neural network; logit-artificial neural network ensemble 1. INTRODUCTION Mode choice modeling is a very important component of transportation planning process. The travelers’ choice or preference for particular transportation mode affects the transportation policies, regulations, and projects. Some examples of the transportation-related issues affected by mode choice decisions include public transport fare and service, parking regulations, and feasibility of new transportation services [1]. Mode choice modeling techniques can be broadly divided in to two categories: statistical and artifi- cial intelligence (AI). Logit model is a popular statistical technique for mode choice modeling. On the other hand, artificial neural network (ANN), among AI techniques, has been used lately for modeling travelers’ mode choice. A detailed comparison between these techniques is given by Karlaftis and Vlahogianni [2]. The comparison of logit model and ANN models in the context of choice modeling can be reviewed in Hensher and Ton, and Agrawal and Schorling [3, 4]. Hensher and Ton have pre- sented the comparison specifically for performance of these methods for commuter mode choice. They concluded in their study that ANN models showed higher performance for predicting individual mode choice predictions. They also pointed out that ANN models have good generalization capabilities and can be developed without specifying the relationship between input and output parameters. However, ANN models are also reported to have over-fitting problem, which means that they give significantly higher accuracies for the training dataset as compared with test dataset [5, 6]. This prob- lem often occurs because of insufficient or lack of validation of the developed model; therefore, *Correspondence to: Uneb Gazder, Department of Civil and Environmental Engineering, King Fahd University of Petro- leum and Minerals, Dhahran 31261, Saudi Arabia. E-mail: unebgazdar@gmail.com Copyright © 2015 John Wiley & Sons, Ltd. JOURNAL OF ADVANCED TRANSPORTATION J. Adv. Transp. 2015; 49:855–866 Published online 24 February 2015 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/atr.1306