Study on a Decision Support System for Large

Download Report

Transcript Study on a Decision Support System for Large

Study on Development and Application of
MAS for Impact Analysis of Large-scale
Shopping Center Development
ZJ. Shen, M. Kawakami, P. Chen
Kanazawa University, Japan
2006. DDSS
Contents
Introduction
Location Regulations for B-shops
Framework of Shopsim-MAS
Policy Scenarios Evaluation
Discussion and Further Research
Background
The commercial environment of many local cities in Japan is
experiencing decline in their centre areas.
Local governments have developed all kinds of city center
generation policies to constrain this trend and revitalize the
central city commercial environment.
It is difficult to evaluate the potential impact of current policies
on the city future due to the uncertainty inherent in urban
system.
MAS simulation is reconized as a tool to visualize impact of
planning policies for presenting the complexity of the urban
system.
Introduction
Picture of center area in metropolitan
prosperous
commercial street
in Osaka
Introduction
Pictures of center area in local city
Decline of commercial environment
Introduction
Pictures of large-shopping mall
In suburban area
Contents
Introduction
Location Regulations for B-shops
Framework of Shopsim-MAS
Policy Scenarios Evaluation
Discussion and Further Research
Location Regulations for B-shops
Planning
regulations on
location of largescale shopping
center
(B-shop)
Urban planning area
Land use zone
Urbanization
•1st low-rise exclusive residential district
•2nd low-rise exclusive residential district
•1st mid-high rise exclusive residential district
•2nd mid-high rise exclusive residential district
•1st residential district
•Exclusive industrial distinct
Promoting Area
Permitting state X
•Commercial district
•Quasi –industrial district
•Industrial distinct
•2nd residential district
•Quasi-residential district
•Neighborhood commercial district
Urbanization
Control Area
White Land
X
B-shops are not permitted to locate in these land zoning district
B-shops can be permitted to locate in these land zoning district
In principle, any development are prohibit in Urbanization Control Area
Location Regulations for B-shops
Land use zoning
Location Regulations for B-shops
Bylaw
regarding
location
and floor
space
Location Regulations for B-shops
Principle scenarios for shopping center location as Decision table for
developer agent (Bylaw regulations and land use zonings)
C1 Scenario
Location is in
C2 center
Location matches
C3 zone type
Location is
consistent with
C4 bylaw
Distance to
existing shops is
C5 appropriate
Set potential
A1 location
Set floor space
A2 up-limit
Center Activation
Y
N
Y
N
/
/
N
/
/
/
Y
Y
Railway Station Development
N
N
Y
Y
Y
Neighborhood Commerce Promoting
N
/
N
/
/
Y
Y
N
N
/
Y
N
/
/
/
/
Y
N
/
/
/
Y
N
/
/
Y
/
/
/
/
/
Y
/
/
/
/
Y
/
/
/
/
R4
/
R5
/
R6
N
R1
/ /
R2 R3
10000 / /
/
/
R7 R8 R9 R10 R11
3000
/
/
R12 R13 R14
/
R15
Location Regulations for B-shops
The location alternatives are limited in the possible areas
according to land use zonings regulation and bylaw regarding
large-scale shopping mall.
These location alternatives reflect the different scenarios of
commercial development.
Contents
Introduction
Location Regulations for B-shops
Framework of Shopsim-MAS
Policy Scenarios Evaluation
Discussion and Further Research
Framework of Shopsim-MAS
provincial city of Japan
 Mono central




> urban sprawl
> suburb house development
> large suburban shopping mall
Poly central


> declination in inner city
> Policy change -> location
regulations of lager shopping mall
Framework of Shopsim-MAS
--- Shop choice (percolation model)
Percolation model for getting
spatial pattern
Percolation probability Ps or Pb
Pb for shopping in B-shop
Ps for shopping in S-shop
Pb + Ps = 1
If Pb > 0.5 then percolation
phenomenon will occur.
To keep the S_shop market
share, Ps should be more than
0.5
S-shop
B-shop
Framework of Shopsim-MAS
--- Shop choice (percolation model)
A random utility model for
shopping Probability in
Percolation model
Agents’ (Household) shop
choice of B_shop or S_shop


S-shop
Chose B_shop if Uib > Uis
Chose S_shop if Uis > Uib
U ij  Vij   ij
B-shop
Framework of Shopsim-MAS
----Shop choice (percolation model)
A random utility model for
household shop choice:
The random factor can be
used to adjust percolation
probability, which will
generate diverse spatial
patterns
U ij  Vij   ij
V
ij

k
f (X
j
kij
)  TC ij
TC      ( xi  x j)  ( y i  y j)
2
ij
i
2
Framework of Shopsim-MAS
----shop choice model
According to local regulations of large scale shopping mall,
influence factors of percolation probability should be set as location
(set as decision table )and floor area.
Xkij is the kth attribute describing store j attracting household i.,
 price: X1j =Pj and
 floor space: X2j =Sj (Price Pj is added by authors)
 distance: Cij is a measure of the disutility of travel between site of
household i and site of shop j. (Cij is added by authors)
Shopping choice in simulation based on utility is deterministic
process,  ij as random factor to control individual choice.
Framework of Shopsim-MAS
--- Shop choice (percolation model)
Cij is a measure of
the disutility of travel
between site of
household i and site
of shop j.
Percolation probability
become unstable in
this case, however it
is relative to its spatial
position.
V
ij

k
f (X
j
kij
)  TC ij
S-shop
H → B-shop
distance
( xh  xB )  ( yh  yB )
H
H → S-shop distance
( xh  xS )  ( yh  yS )
B-shop
Framework of Shopsim-MAS
--- Shop choice (percolation model)
percolation probability is
shifted if household
position is near ot far away
from a shop.
To S-shop
Cij  shorter
Ps(Sj,Pj,Cij)  larger
H
S-shop
H
To S-shop
B-shop
Cij  longer
Ps(Sj,Pj,Cij)  smaller
Framework of Shopsim-MAS
----spatial pattern (percolation model)
random value was set to
translate the probability of
random utility model into
simulation
in Uib, 10000
in Uis, 10000
Rate of shoping
in B_shop=0.08
in S_shop=0.92
price in S 300, in B 200
travel cost 120/cell
Framework of Shopsim-MAS
----spatial pattern (percolation model)
random value was set to
translate the probability of
random utility model into
simulation
in Uib, 10000
in Uis, 5000
Rate of shoping
in B_shop=0.5
in S_shop=0.5
price in S 300, in B 200
travel cost 120/cell
Framework of Shopsim-MAS
----spatial pattern (percolation model)
random value was set to
translate the probability of
random utility model into
simulation
in Uib, 5000
in Uis, 5000
Rate of shopping
in B_shop=0.94
in S_shop=0.06
price in S 300, in B 200
travel cost 120/cell
Framework of Shopsim-MAS
----spatial pattern (percolation model)
random value was set to
translate the probability of
random utility model into
simulation
in Uib, 2000
in Uis, 2000
Rate of shopping
in B_shop=0.16
in S_shop=0.84
price in S 300, in B 200
travel cost 120/cell
Framework of Shopsim-MAS
----spatial pattern (percolation model)
random value was set to
translate the probability of
random utility model into
simulation
in Uib, 500
in Uis, 500
(critical point)
Rate of shoping
in B_shop=0.24
in S_shop=0.76
price in S 300, in B 200
travel cost 120/cell
Framework of Shopsim-MAS
---shop choice model
Therefore, percolation Probability of B_shop or S_shop is
decided by Pj, Sj and Cij. For translating probability of random
utility model into agent’s individual behavior, a random variable
is defined.
If percolation probability changed gradually, the spatial pattern
of percolation will be changed gradually. This phenomenon can
be used in the market share simulation using MAS.
However, how about fitness of Individual shopping choice based
on ramdam utility and percolation probability in simulation is still
a further study.
Framework of Shopsim-MAS
----uban space and agents
Urban Space
Agents




Planner
Developer
Shop
 S-shop
 B-shop
Household
Framework of Shopsim-MAS
----object model
Framework of Shopsim-MAS
----Simulation Process
1.
2.
3.
4.
5.
The user of Shopsim-MAS defines a policy scenario to be
implemented. -> decision table
The planner agent sets the spatial structure and initiates the
scenario.
S-shop agents and existing B-shop agents are created in the
urban space. Household agents are created and distributed to
the whole central city urban planning area.The developer agent
places the new B-shop in urban space according to defined
scenarios.
The user sets the initial values of parameters .For clear
simulation results, random value is set as 500 under critical
point.
Households then decide where to go shopping until their
demands are fulfilled (demands of each household=50).
Contents
Introduction
Location Regulations for B-shops
Framework of Shopsim-MAS
Policy Scenarios Evaluation
Discussion and Further Research
Policy Scenarios Evaluation
----Define four cases of scenario
Base Scenario
No new B-shop are permitted to develop
Centre Activation(CA)
B-shop can only locate in the centre commercial area without
upper limitation for floor space.
Railway Station Development (RSD)
B-shop can only be opened near the station, with an upper
limitation of 10000 m2.
Neighbouring Commerce Promotion (NCP)
B-shop can only locate in neighbour commercial area, with an
upper limitation of 3000 m2.
Policy Scenarios Evaluation
----Analysis
of
CA
scenario
Floor space =
3000m2
5000m2
Market share of
existing B-shop
8000m2
10000m2
Market share of
the new B-shop
15000m2
20000m2
Market share
of S-shop
The spatial effects of CA scenario as
shown in figures.
Sale statistics in CA scenario
It can be see that CA scenario do have
effect in improving the market
performance of the city centre, but may do
severe harm to the centre S-shops at the
same time if there is no limitation on Bshop’s scale.
Policy Scenarios Evaluation
----Compare scenarios
To compare three scenarios,
the floor space of the new Bshop is set same as 3000m2.
 Both Figures show that in RSD
and NCP scenario, the loss of
market share of S-shops caused
by the new B-shop is more than
in CA scenario .
Both Figures also indicate that
CA scenario might be the only
effective to improve center
commerce among three
scenarios.
Sale statistics of Center shops
and S-shops
Base
CA
Market share of existing
B-shop
RSD
NCP
Market share of Sshop
Contents
Introduction
Location Regulations for B-shops
Framework of Shopsim-MAS
Policy Scenarios Evaluation
Discussion and Further Research
Discussion and Further Research
The use of MAS for impact analysis of large scale shopping
center development regulations is proposed in this paper.
By introducing real urban land use zoning to form agent’s
behavior constraints, the Shopsim-MAS simulate the virtual
urban space in a more practical way in the context of urban
planning.
Percolation model and random utility model are employed in this
simulation and spatial pattern of the market share influenced by
urban bylaw and planning regulations can be visualized.
The simulation results of four possible policy scenarios indicate
that to develop new B-shop in the city center might be an
effective measure to improve commercial activity of city centre.
However, how about the behavior of households (random
factors in this simulation that will influence the spatial pattern of
market share) ?