Page 1
ENTER 2015 Research Track Slide Number 1
Spatiotemporal Analysis of Rambling Activities:
Approach to Inferring Visitor Satisfaction
Masakatsu Ohtaa
Yuta Watanabeb
Toshiaki Miyazakib
aNTT Network Innovation Laboratories,
NTT Corporation, Japan
bSchool of Computer Science and Engineering,
The University of Aizu, Japan
Page 2
ENTER 2015 Research Track Slide Number 2
Challenge
which?
trajectory of trip
Page 3
ENTER 2015 Research Track Slide Number 3
Agenda
1. Introduction
2. Approach– Spatiotemporal analysis
1. Case Study– Campus festival
1. Conclusions
Page 4
ENTER 2015 Research Track Slide Number 4
Planning with ICT
• Make practical plans under time & budget constraints
Page 5
ENTER 2015 Research Track Slide Number 5
Rambling Activities
stop by
stop by
original
• Trajectory deviates from efficient route
actual
Page 6
ENTER 2015 Research Track Slide Number 6
Encounter the Unexpected
what?Interesting!
bicycle?
clog?
man?
Page 7
ENTER 2015 Research Track Slide Number 7
Independent Choiceswhere to go next?
after determining
vs.
Expectation increasescontent about
choice
Page 8
ENTER 2015 Research Track Slide Number 8
Rambling Activities and Visitor Satisfaction
• Rambling activities meet following conditions:– Encounter the unexpected– Independent Choices
Person rambling in area is likely attracted to it
Page 9
ENTER 2015 Research Track Slide Number 9
Sustainable Development
inducerambling
urban plannersevent organizers …
Page 10
ENTER 2015 Research Track Slide Number 10
Goal• Contributes to check phase
‒ Visitors were rambling in area?‒ Area is attractive to them?
Page 11
ENTER 2015 Research Track Slide Number 11
Agenda
1. Introduction
2. Approach– Spatiotemporal analysis
1. Case Study– Campus festival
1. Conclusions
Page 12
ENTER 2015 Research Track Slide Number 12
Prominent Attributes
• Spatiotemporal Dispersal of Visits– Many visited spots– Various dwell times at various spots
• Unplanned “Stopping by”– Not efficient route– Many intersections in trajectory
Page 13
ENTER 2015 Research Track Slide Number 13
Spatiotemporal Analysis
3D curve: locations of visited spotsxyz dwell times at visited spots
more rambling
Y
X
Z
Y
X
Z
Page 14
ENTER 2015 Research Track Slide Number 14
Knot Theory
unknot 3.1 4.1 5.1 5.2
6.1 6.2 6.3 7.1 7.2
7.3 7.4 7.5 7.6 7.7
distinct knots (examples)
equivalent knots
unknot
Page 15
ENTER 2015 Research Track Slide Number 15
Creating Knot
3D curve3D curve
ClosingClosing
SimplificationSimplification
start end
dwell time
trajectory
deform to equivalent knots
3D curve
springspring
start
beadbead
spot spot
spot end
Page 16
ENTER 2015 Research Track Slide Number 16
ExamplesDispersal & Unplanned Dispersal & Unplanned
Spatially biasedSpatially biased
Temporally biasedTemporally biased
Well scheduledWell scheduled
simplify
knot
unknot
unknot
unknotdwell time length
transform
Page 17
ENTER 2015 Research Track Slide Number 17
Determination of Rambling
knot
unknotdwell time length
transformRambling
NOT Rambling
trajectory
Page 18
ENTER 2015 Research Track Slide Number 18
Agenda
1. Introduction
2. Approach– Spatiotemporal analysis
1. Case Study– Campus festival
1. Conclusions
Page 19
ENTER 2015 Research Track Slide Number 19
Campus Festival
Similar to towns• Open to public• Various spots and events
food stands, concerts, lectures…
Restricted space• Controlled experiment• Many trajectories at a time
Date: Oct. 12th and 13th, 2013Location: The University of Aizu Visitors: 5,200
Date: Oct. 12th and 13th, 2013Location: The University of Aizu Visitors: 5,200
Page 20
ENTER 2015 Research Track Slide Number 20
Experiment Design
Indoor
Outdoor
100 mbeacon reader( spot )
beacon venue
Date: Oct. 13th, 2013Number of spots: 21Participants: 135 groupsRegular visitors: 77%
Date: Oct. 13th, 2013Number of spots: 21Participants: 135 groupsRegular visitors: 77%
Page 21
ENTER 2015 Research Track Slide Number 21
Evaluations
1. Inferring visitor satisfaction– Were participants who rambled
satisfied?
1. Regular and non-regular difference– Did non-regular visitors go around
venue much more?
Page 22
ENTER 2015 Research Track Slide Number 22
Satisfaction Measure
• Difficult to understand visitor’s true impressions• New measure:
• If overspending > 1.25:
more satisfied than expected
from 0.75 to 1.25: stayed on scheduled(estimated error in planning: 0.25)
overspending = actual spending time
planned spending time
Page 23
ENTER 2015 Research Track Slide Number 23
1) Visitor Satisfaction
• Significant differences (Fisher’s Exact Test: p=0.008)
• Finding‒ Detecting rambling activities‒ Inferring visitor satisfaction
• Significant differences (Fisher’s Exact Test: p=0.008)
• Finding‒ Detecting rambling activities‒ Inferring visitor satisfaction
more satisfied with festival than originally expected
more satisfied with festival than originally expected
Page 24
ENTER 2015 Research Track Slide Number 24
2) Regular and non-Regular Visitors
• NO significant differences (Fisher’s Exact Test: p=0.51)
• Finding‒ Influence of Social Validation:
Imitate others’ opinions and behaviors
• NO significant differences (Fisher’s Exact Test: p=0.51)
• Finding‒ Influence of Social Validation:
Imitate others’ opinions and behaviors
*regular visitors: 77%
Page 25
ENTER 2015 Research Track Slide Number 25
Conclusions
• Detected trajectories caused by rambling activities• Inferred visitor satisfaction
Future work:Compare rambling activities between different environments
City A City B
Page 26
ENTER 2015 Research Track Slide Number 26
references
Page 27
ENTER 2015 Research Track Slide Number 27
Spot 1Spot 1 Spot 2Spot 2 Spot 3Spot 3 Spot 4Spot 4 Spot 5Spot 5
Adjusted Dwell Time
real dwell time
adjusted dwell time
order of length
10 120 1030 60
1 4 12 3
1L 4L 1L2L 3L
trajectory
• Emphasize variation in length of dwell time
*L: parameter
Page 28
ENTER 2015 Research Track Slide Number 28
Conventional Method
overspending time rateoverspending time rate
Difficult to infer participant satisfaction by “trajectory length” and “spending-time at venue”
more satisfied with festival than originally expected