Introduction of “People flow project” --Understanding of dynamic change about
people in the city
Yoshihide Sekimoto,
Project Associate Prof.
Center for Spatial Information Science,
The University of Tokyo
1
Center for Spatial Information Science,
the University of Tokyo
History of CSIS
1988
“National cartography museum”
Recommendation by the Science
Council of Japan
Launch of Joint research
program using spatial
data platform
1996
National committee for a
research center for GIS
Enhancement of
Center functions
as joint facility
Spatial data sharing system
Spatial data clearinghouse
CSIS Catalogue service
Address matching service
1998
CSIS established
2005
Moved to new Kashiwa campus
History
1998
Academic Portal of GIS
GISSchool
Design studio for GI
2006
Re-launch of CSIS as a National
(inter-university) joint-research facility
Spatial data platform
for joint research
2007
Needs of People flow data
• Needs of time-based location information
of many people are increasing…
Prevention of secondary disaster
in the complex urban space
Big earthquake
Flood disaster to
underground mall
Stimulation of the economy
by people gathering
Outdoor advertising
Events like Festival
People flow project in CSIS
• People flow project since 2008 in CSIS introduces
some technologies and results about people flow
http://pflow.csis.u-tokyo.ac.jp
4
Existing our research
Sekimoto et al. (2011)* had proposed reconstruction
method using large-scale fragmentary social survey data
* Y. Sekimoto et al. PFLOW: Reconstruction of people flow by recycling large-scale fragmentary
social survey data, IEEE Pervasive Computing, Vol.10(4) pp.27-35, 2011.
Reconstruction of macroscopic people flow in Central
Tokyo using person trip (PT) survey data
5
3D visualization
3D visualization with 1-km2 mesh
6
Data source: questionnaire
(e.g. Person trip data)
Place staying at 3:00 in the morning
1st place to go
2nd place to go
Home
Office or School
Other places (Rough address)
Other places (Place name)
Kind of place
Trip
Departure time
Arrival time
Trip
object
Purpose
Sub-trip
Transportation mode
Travel time
Transfer point
Main part of PT survey sheet (from the Tokyo Metropolitan Region Transportation Planning
Commission web site "http://www.tokyo-pt.jp/data/file/tebiki.pdf")
7
Spatio-temporal interpolation from OD data
Station
(11:45)
Sub trip 2
(Railway)
Home
(11:42)
11:42 (Lat1,Lon1)
11:43 (Lat2,Lon2)
11:44 (Lat3,Lon3)
11:45 (Lat4,Lon4)
11:46 (Lat5,Lon5)
11:47 (Lat6,Lon6)
11:48 (Lat7,Lon7)
11:49 (Lat8,Lon8)
11:50 (Lat9,Lon9)
11:51 (Lat10,Lon10)
11:52 (Lat11,Lon11)
11:53 (Lat12,Lon12)
11:54 (Lat13,Lon13)
Sub trip 1
(Walking)
• XXXX
Office
(11:54)
Station
(11:49)
Sub trip 3
(Walking)
a) Geocoded OD of
(a) each sub trip
Railway
geometry
(Green line)
Railway rough route
(Blue line)
Walking
rough route
(Red line)
Road
topology
(Gray line)
Road
geometry
(Gray line)
c)(c)
Interpolation at each 1 minute-intervals
Railway
topology
(Green line)
b) Route choice along
road/railway topology
(b)
8
Reconstruction accuracy
9
JICA-PT data （Hanoi）
10
JICA-PT data（Manila）
11
Project structure
Activity model
of each people
Observation data
PT data
Estimation
Observe
・・・・
Disaggregated
moving model
Aggregated
distribution
＋
People flow in real world
12
Classification of each data source
Quality of sample
PT data
Real-time
property
13
Data source: Twitter
2,500
Tweet per hour
1時間あたりtweet数
Daily fluctuation
for two weeks
2,000
1,500
1,000
500
0
0
2
4
0:00
3:00
6:00
9:00
12:00
15:00
18:00
21:00
6
8
10 12 14 16 18 20 22
時刻
Time
From Dr. Fujita in CSIS
14
Data source: Four square
Four square
mapping data of
one person for
two years in SF
http://www.weeplaces.com
15
Data source: mobile phone base station
Mobile Spatial
Statistics
Operational data
De-identification
Privilege
Aggregation
Aggregated
population
Mobile Spatial Statistics（From NTTDocomo web site:
http://www.nttdocomo.co.jp/corporate/disclosure/mobile_spatial_statistics/）
16
Data source: mobile phone GPS data
Density map from Auto-GPS data
（ZENRIN DataCom CO.,LTD. http://lab.its-mo.com/densitymap/）
17
Data assimilation technology
using observation data
• Data assimilation is integration of model and observation
data (based on e.g. Recursive Baysian Estimation…)
RMSE of the number of people
Red: No assimulation
Blue: Assimulation
Time(hour)
Total RMSE of stations between
observed and estimated data
Total RMSE of roads between
observed and estimated data
18
Global approach
(JICA-PT data + OpenStreetMap)
City
Population
(million)
Survey
year
Number of
people
Number of
trips
Manila
Kuala Lumpur
Damascus
Managua
Bucharest
Phnom Penh
Chengdu
Belem
Jakarta
Tripoli
9.45
1.39
3.08
1.20
2.15
1.15
3.09
1.78
2.10
0.33
1996
1997
1998
1998
1998
2000
2000
2000
2000
2001
231,889
80,560
38,490
24,854
67,509
18,664
31,188
24,043
423,237
3,608
471,035
218,460
81,698
54,138
143,311
40,369
70,199
59,529
1,083,280
7,615
…
(a) Belem
…
…
(b) Ho Chi Minh
…
…
Ratios of various modes of
main transportation
(2-wheeler/car/taxi/bus/rail)
2%/10%/25%/58%/4%
29%/44%/2%/23%/2%
4%/25%/15%/56%/0%
2%/25%/4%/69%/0%
0%/19%/0%/27%/54%
89%/11%*1/-/81%/10%/4%/4%/0%
15%/13%/2%/70%/0%
2%/0%/42%/56%/0%
35%/29%/17%/19%/0%
…
(c) Chengdu
19
Many joint researches through
“People Flow Data Set”
【Transportation】
# Research on improving the efficiency of urban transport systems using portable personal
mobility.(iTransport Lab, Ltd.)
# A simulation of tourist flow patterns in the Sendai metropolitan area using the People Flow
Analysis Platform. Masayoshi Tanishita (Chuo University)
# Utilization of statistical data in urban transport planning. (Ritsumeikan Asia Pacific University
Department of Asian Pacific Studies)
【spatio-temporal analysis】
# Detection of patterns in travel routes using position information and travel times (Kobe University
Graduate School of Engineering)
# Development of a spatio-temporal data model for analysis of spatio-temporal behavior using GIS.
(Tokyo Metropolitan University)
【Risk analysis】
# A model for the transmission of novel infectious diseases.
(University of Tokyo Institute of Industrial Science)
# An investigation of of disaster risk using GIS. (Aichi Institute of Technology Department of
Environmental Engineering)
【Personal information and security】
# On the anonymization of personal information and its two-dimensional use (Information Grand
Voyage Project). (Mitsubishi Research Institute, Inc.)
【Environment】
# Development of a scenario for fine spatial output and changes in land use via unified system
analysis. (National Institute for Environmental Studies)
【Marketing】
# A study of consumer respiration models using person-trip data. (Fine Analysis, LLC)
20
Influenza
Day 3
Day 27
Day 46
Aihara & Suzuki lab in IIS, Univ. of Tokyo
21
Thank you !
sekimoto@csis.u-tokyo.ac.jp
22