Street choice logit model for visitors in shopping districts

Ko Kawada, Takashi Yamada, Tatsuya Kishimoto

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

In this study, we propose two models for predicting people’s activity. The first model is the pedestrian distribution prediction (or postdiction) model by multiple regression analysis using space syntax indices of urban fabric and people distribution data obtained from a field survey. The second model is a street choice model for visitors using multinomial logit model. We performed a questionnaire survey on the field to investigate the strolling routes of 46 visitors and obtained a total of 1211 street choices in their routes. We proposed a utility function, sum of weighted space syntax indices, and other indices, and estimated the parameters for weights on the basis of maximum likelihood. These models consider both street networks, distance from destination, direction of the street choice and other spatial compositions (numbers of pedestrians, cars, shops, and elevation). The first model explains the characteristics of the street where many people tend to walk or stay. The second model explains the mechanism underlying the street choice of visitors and clarifies the differences in the weights of street choice parameters among the various attributes, such as gender, existence of destinations, number of people, etc. For all the attributes considered, the influences of DISTANCE and DIRECTION are strong. On the other hand, the influences of Int.V, SHOPS, CARS, ELEVATION, and WIDTH are different for each attribute. People with defined destinations tend to choose streets that ―have more shops, and are wider and lower. In contrast, people with undefined destinations tend to choose streets of high Int.V. The choice of males is affected by Int.V, SHOPS, WIDTH (positive) and CARS (negative). Females prefer streets that have many shops, and couples tend to choose downhill streets. The behavior of individual persons is affected by all variables. The behavior of people visiting in groups is affected by SHOP and WIDTH (positive).

Original languageEnglish
Pages (from-to)154-166
Number of pages13
JournalBehavioral Sciences
Volume4
Issue number3
DOIs
Publication statusPublished - 2014 Sep 1

Fingerprint

logit analysis
Logistic Models
district
Homeless Persons
Weights and Measures
Regression Analysis
pedestrian
syntax
utility functions
Pedestrians
Surveys and Questionnaires
questionnaire survey
field survey
multiple regression
regression analysis
questionnaires
automobile
gender
prediction
Direction compound

Keywords

  • Logit model
  • Pedestrian behavior
  • Pedestrian distribution
  • Route choice
  • Shopping district
  • Space syntax
  • Street choice
  • Strolling visitors

ASJC Scopus subject areas

  • Psychology(all)
  • Ecology, Evolution, Behavior and Systematics
  • Behavioral Neuroscience
  • Genetics
  • Development

Cite this

Street choice logit model for visitors in shopping districts. / Kawada, Ko; Yamada, Takashi; Kishimoto, Tatsuya.

In: Behavioral Sciences, Vol. 4, No. 3, 01.09.2014, p. 154-166.

Research output: Contribution to journalArticle

Kawada, Ko ; Yamada, Takashi ; Kishimoto, Tatsuya. / Street choice logit model for visitors in shopping districts. In: Behavioral Sciences. 2014 ; Vol. 4, No. 3. pp. 154-166.
@article{6fa0d9c4e1bc4dfbb201d9a94b69a831,
title = "Street choice logit model for visitors in shopping districts",
abstract = "In this study, we propose two models for predicting people’s activity. The first model is the pedestrian distribution prediction (or postdiction) model by multiple regression analysis using space syntax indices of urban fabric and people distribution data obtained from a field survey. The second model is a street choice model for visitors using multinomial logit model. We performed a questionnaire survey on the field to investigate the strolling routes of 46 visitors and obtained a total of 1211 street choices in their routes. We proposed a utility function, sum of weighted space syntax indices, and other indices, and estimated the parameters for weights on the basis of maximum likelihood. These models consider both street networks, distance from destination, direction of the street choice and other spatial compositions (numbers of pedestrians, cars, shops, and elevation). The first model explains the characteristics of the street where many people tend to walk or stay. The second model explains the mechanism underlying the street choice of visitors and clarifies the differences in the weights of street choice parameters among the various attributes, such as gender, existence of destinations, number of people, etc. For all the attributes considered, the influences of DISTANCE and DIRECTION are strong. On the other hand, the influences of Int.V, SHOPS, CARS, ELEVATION, and WIDTH are different for each attribute. People with defined destinations tend to choose streets that ―have more shops, and are wider and lower. In contrast, people with undefined destinations tend to choose streets of high Int.V. The choice of males is affected by Int.V, SHOPS, WIDTH (positive) and CARS (negative). Females prefer streets that have many shops, and couples tend to choose downhill streets. The behavior of individual persons is affected by all variables. The behavior of people visiting in groups is affected by SHOP and WIDTH (positive).",
keywords = "Logit model, Pedestrian behavior, Pedestrian distribution, Route choice, Shopping district, Space syntax, Street choice, Strolling visitors",
author = "Ko Kawada and Takashi Yamada and Tatsuya Kishimoto",
year = "2014",
month = "9",
day = "1",
doi = "10.3390/bs4030154",
language = "English",
volume = "4",
pages = "154--166",
journal = "Behavioral Sciences",
issn = "2076-328X",
publisher = "Multidisciplinary Digital Publishing Institute",
number = "3",

}

TY - JOUR

T1 - Street choice logit model for visitors in shopping districts

AU - Kawada, Ko

AU - Yamada, Takashi

AU - Kishimoto, Tatsuya

PY - 2014/9/1

Y1 - 2014/9/1

N2 - In this study, we propose two models for predicting people’s activity. The first model is the pedestrian distribution prediction (or postdiction) model by multiple regression analysis using space syntax indices of urban fabric and people distribution data obtained from a field survey. The second model is a street choice model for visitors using multinomial logit model. We performed a questionnaire survey on the field to investigate the strolling routes of 46 visitors and obtained a total of 1211 street choices in their routes. We proposed a utility function, sum of weighted space syntax indices, and other indices, and estimated the parameters for weights on the basis of maximum likelihood. These models consider both street networks, distance from destination, direction of the street choice and other spatial compositions (numbers of pedestrians, cars, shops, and elevation). The first model explains the characteristics of the street where many people tend to walk or stay. The second model explains the mechanism underlying the street choice of visitors and clarifies the differences in the weights of street choice parameters among the various attributes, such as gender, existence of destinations, number of people, etc. For all the attributes considered, the influences of DISTANCE and DIRECTION are strong. On the other hand, the influences of Int.V, SHOPS, CARS, ELEVATION, and WIDTH are different for each attribute. People with defined destinations tend to choose streets that ―have more shops, and are wider and lower. In contrast, people with undefined destinations tend to choose streets of high Int.V. The choice of males is affected by Int.V, SHOPS, WIDTH (positive) and CARS (negative). Females prefer streets that have many shops, and couples tend to choose downhill streets. The behavior of individual persons is affected by all variables. The behavior of people visiting in groups is affected by SHOP and WIDTH (positive).

AB - In this study, we propose two models for predicting people’s activity. The first model is the pedestrian distribution prediction (or postdiction) model by multiple regression analysis using space syntax indices of urban fabric and people distribution data obtained from a field survey. The second model is a street choice model for visitors using multinomial logit model. We performed a questionnaire survey on the field to investigate the strolling routes of 46 visitors and obtained a total of 1211 street choices in their routes. We proposed a utility function, sum of weighted space syntax indices, and other indices, and estimated the parameters for weights on the basis of maximum likelihood. These models consider both street networks, distance from destination, direction of the street choice and other spatial compositions (numbers of pedestrians, cars, shops, and elevation). The first model explains the characteristics of the street where many people tend to walk or stay. The second model explains the mechanism underlying the street choice of visitors and clarifies the differences in the weights of street choice parameters among the various attributes, such as gender, existence of destinations, number of people, etc. For all the attributes considered, the influences of DISTANCE and DIRECTION are strong. On the other hand, the influences of Int.V, SHOPS, CARS, ELEVATION, and WIDTH are different for each attribute. People with defined destinations tend to choose streets that ―have more shops, and are wider and lower. In contrast, people with undefined destinations tend to choose streets of high Int.V. The choice of males is affected by Int.V, SHOPS, WIDTH (positive) and CARS (negative). Females prefer streets that have many shops, and couples tend to choose downhill streets. The behavior of individual persons is affected by all variables. The behavior of people visiting in groups is affected by SHOP and WIDTH (positive).

KW - Logit model

KW - Pedestrian behavior

KW - Pedestrian distribution

KW - Route choice

KW - Shopping district

KW - Space syntax

KW - Street choice

KW - Strolling visitors

UR - http://www.scopus.com/inward/record.url?scp=84987918606&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84987918606&partnerID=8YFLogxK

U2 - 10.3390/bs4030154

DO - 10.3390/bs4030154

M3 - Article

VL - 4

SP - 154

EP - 166

JO - Behavioral Sciences

JF - Behavioral Sciences

SN - 2076-328X

IS - 3

ER -