No.41 In 2019
Title
Modeling cyclists' facility choice and its application in bike lane usage forecasting
Recipient
Nguyen Duc-Nghiem, Nguyen Hoang-Tung, Aya Kojima, Hisashi Kubota
Reason for award
The availability of new travel spaces, such as bike lanes, does not necessarily mean that bikers will simply change their routes. Rather, it is likely that many bikers will continue to ride on sidewalks as before.
In response to this expectation, the authors developed a model that predicts where bikers will choose to ride: on the sidewalk, in the car lane of the road, near the curb of the road, or in a new bike lane on the road. The model is a binomial logistic regression model based on data observed on fifteen roads in Saitama City. One distinctive feature of the model as the analytical method is that it uses Bayesian Model Averaging, which has attracted significant attention in recent years; because explanatory variables that have little influence according to careful verification have been removed, the model is simpler, easier to handle, and more predictive in nature.
For the final model, nine of the eighteen variables set at the time of the study were selected. Some of the nine variables indicate that road width, bike type (racing or mountain bike), roadside strip width, and new bike lane width were selected as variables that influenced bikers' choice to ride on the road, and rest of them indicate that the effective width of sidewalks, the presence of bus stops, the presence of parked vehicles, gender, availability of seats for children, and whether or not bikers are biking as a group are each included in the model as variables that influence sidewalk riding choices. In particular, the presence of bus stops, the effective width of sidewalks, and bike type have been found to have significant influences. The model the authors built has been validated with data that shows both before and after the bike lanes were introduced, and it has been found to be very effective for explanatory purposes. That has made it possible to predict how new bike lanes will be used. Although care must be taken when applying the methodology to other cities, it provides a useful methodology for the preliminary estimation of project effects and post-project evaluation.
With the issuance of the Guidelines for Creating a Safe and Comfortable Bicycle Environment in 2012, the rule that bikes should, as a general principle, be ridden on the roadway has been reaffirmed in Japan, and the installation of bike lanes within the road space has been vigorously promoted throughout the country, but it is important to prepare bike spaces where bikes can ride safely within the limited road space in consideration of pedestrians, and to have bikers use them. This paper is the result of a field survey on the subject, and the data has been used to construct a model that can be applied in practical terms by incorporating new, superior methods of mathematical statistics together with knowledge of civil engineering and urban engineering; the usefulness of the model has been verified using pre- and post-survey data. The authors have achieved excellent results using academic, practical, and cross-disciplinary approaches that contribute to bringing about a safe and comfortable mobility society.
Overview of the Paper
On roads with new bike lanes, not all bikers will use them in all places; they may continue to use sidewalks as in the past. With the recent development of bike lanes in many parts of Japan, predicting where bikers will choose to ride has become an important issue. In this study, a model was developed to predict how bikers will make this choice. A Bayesian model average method was used to select the best model based on the results of field surveys in Saitama City, to identify the factors that influence the prediction of where bikers will choose to ride. These findings will be useful to planners and designers in the development of new bike lanes.