ARC GIS 10.1: Answers The Length Of The Proxy Of Travel

To estimate impedance function, distance as a proxy of travel cost between pairs of origins and destinations is calculated by ARC GIS 10.2 which considers the length of the shortest available route. To calibrate the best impedance function, the frequency distribution of walking trips is plotted against distance. By fitting different types of functions (i.e., exponential, multiplicative exponential, linear, logarithmic, polynomial, power) and comparing standard error of the estimates’ values, the best impedance function is derived which is negative exponential (see Table 2 and Figure 3). It is worth noting that, while it is less discussed that which of different impedance functions’ specifications contain the supreme effect among accessibility
While environmental variables in terms of both origin and destination are considered in modeling, only destination variables are significantly appeared in the model. The destination choice model shows that people tend to walk to destinations which are closer and also benefit from more connected network as well as various land uses; these characteristics at destinations increase the probability of choosing that destinations to walk to. Figure 5 shows the utility-based accessibility measure for each TAZ using equation
It is worth noting that in context of discrete choice models, the Psuedo-R squared value between 0.2 to 0.4 is considered as perfect fit (Colombo et al., 2005). Table 3 reveals that the distance deters pedestrians from walking to destinations which is in consistence with other studies investigating walking behavior (Eash, 1999; Khan et al., 2014; Lee and Moudon, 2006; Sung and Lee, 2015). Furthermore, the model emphasizes on the role of built environments which benefit from diverse land uses (higher values of population density, entropy as well as job-population balance indices) and connected networks (such as higher values of link-connectivity as well as pedestrian catchment area and less dominance of loops and lollipops pattern within street network) which encourage pedestrians to walk to their intended destinations. These findings accord with other studies which recognize land use diversity as a motivating factor for pedestrians to walk (Khan et al., 2014; Lee and Moudon, 2006; Sung and Lee, 2015). Furthermore, as far as we know, the effect of network connectivity is neglected in the previous studies of pedestrian destination choice