AI in Urbanism II xinyu zhang
The Urban Vitality is the urban synergy from “variety" of social , economical opportunities and environment . Good cities tend to be a balance of a reasonably ordered and legible city form.The project’s focus is on the features of social , economy and environment to measure urban vitality and predict feature importance.
Why this matters ?
It could build a design support tool to predict the impact of a design intervention on urban vitality.
What is the specific question?
How to predict the features importance by calculating urban vitality before build a support tool ?
Acce
Environment UV
y sit den ix nt me use m art Ap dingil Bu
ssibi lity
NDVI
NDVI (Normalized difference vegetation index) Apartment density (number/m2) Building-use mix (residential ratio /service ratio /other ratio) Urban fabric density ( streetscapes, buildings, soft and hard landscaping, signage, lighting, roads and other infrastructures surfaced areas range)
URBAN VITALITY
Economy
Social
Accessibility (weight of distance between living sites to bus/metro stops) Habitation density (people number /m2) Commercial density (shop number/m2)
Habitation density Commerical density Service density Urban fabric density
Service density (market number/m2)
NDVI Apartment density Building-use mix Accessibility Habitation density Commercial density Service density Urban fabric density
Algorithm
KNN(K-Nearest Neighbours) classification
Artificial neural network
Environment UV
URBAN VITALITY
Economy
Social
Methodology
[1.0] [0.75] [0.5] [0.25] [0.0]
Data samples (test file)
Assign urban vitality score
Artificial neural network
Trained model
[0 -10 ] [ 1 0 - 20 ] [ 20 - 30 ] [ 30 - 40 ] [ 40 - 50 ] [ 50 - 60 ] [ 60 - 70 ] [ 70 - 80 ] [ 80 - 90 ] [ 90 - 1 00]
Data samples (training file)
Trained model
Predict score
Fill the map
Predict feature importance
Process
Data samples (test file)
Data Sample(test file)
Process
[1.0] [0.75] [0.5] [0.25] [0.0]
Data samples (test file)
Assign urban vitality score
Data Sample(test file)
Score
[1] [1] [1] [1] [0.75] [0.75] [0.75] [0.75] [0.5] [0.5] [0.5] [0.5] [0.25] [0.25] [0.25] [0.25] [0.0] [0.0] [0.0] [0.0]
Process
[1.0] [0.75] [0.5] [0.25] [0.0]
Data samples (test file)
Assign urban vitality score
Artificial neural network
Data Sample(test file)
Trained model
score
Predict index(test file)
Trained model
Predict score and classification
[0 -10 ] [ 1 0 - 20 ] [ 20 - 30 ] [ 30 - 40 ] [ 40 - 50 ] [ 50 - 60 ] [ 60 - 70 ] [ 70 - 80 ] [ 80 - 90 ] [ 90 - 1 00]
Data samples (training file)
Trained model
Predict score
Fill the map
Predict feature importance
LOW urban vitality
0 -10 1 0 - 20 20 - 30 30 - 40 40 - 50 50 - 60 60 - 70 70 - 80 80 - 90
HIGH urban vitality
90 - 1 00 %
[0 -10 ] [ 1 0 - 20 ] [ 20 - 30 ] [ 30 - 40 ] [ 40 - 50 ] [ 50 - 60 ] [ 60 - 70 ] [ 70 - 80 ] [ 80 - 90 ] [ 90 - 1 00]
Data samples (training file)
Trained model
Predict score
Fill the map
a class of techniques for assigning scores to input features to a predictive model that indicates the relative importance of each feature when making a prediction.
Predict feature importance
Building use mix
Apartment density
Habitation density
NDVI
Service density
Commercial density
Accessibility
Urban fabric
Importance
Features impact urban vitality
Variables
Control variable
Actual (Mean)
Predicted
100
urban vitality
80
60
40
20
0
1.05
8.95
2.11 3.16
0
9.47
5
10
14.09
0
25.36
28.17
36.63
20
Urban fabric density
39.44
40
0
0.0005
accessibility
0.001
0.0015
0.002
0
commercial density
0.001
0.002
0.003
service density
100
urban vitality
80
60
40
20
0 0
0.5
NDVI
0
0.5
1
Habitation density
0
0.1
0.2
0.3
Apartment density
0.4
−400
−200
Building use mix
0
High urban vitality space
90 - 1 00 %
Placa de Lesseps Park
Passeig de GrĂ cia Commercial
Low urban vitality space
0- 1 0 %
Transportation railway Transportation
Pedralbes Residential
Parc de la Ciutadella Park
It needs to reset relationship between features and urban vitality on different urban land use. Algorithm
KNN(K-Nearest Neighbours) classification
Industry
Commercial
Transport Residential Nature
Diversity of urban landuse model -k-nearest neighborhood
Start
Input define K
Calculate distance
(test sample and training sample)
class and visualization Sort distance
Take K-nearest neighbours
Apply the simple majority
Diversity of urban landuse model
Start
K1
Input define K
K2
K3
Residential Commercial Nature
K4
K6
Industry
Transport
Diversity of urban landuse model -k-nearest neighborhood
class and visualization Sort distance
Take K-nearest neighbours
Apply the simple majority
K1
K2
K3
K4
K6
Residential
Commercial
Nature
Industry
Transport
urban fabric service density accessibility commercial density
K6
BUM
Industry
urban fabric service density accessibility BUM
service density
commercial density HD NEW AD NEW
NDVI
commercial density
Service density
Commercial density
Habitation density
HD NEW
NDVI
Nature AD NEW
Apartment density
K4 BUM
Habitation density
K3 urban fabric
Commercial density
accessibility
Apartment density
BUM
Building use mix
Commercial
Building use mix
Residential
Urban fabric
K2
Accessibility
Building use mix
Accessibility
importance
importance
K1
Service density
importance
Building use mix
NDVI
Urban fabric
Apartment density
NDVI
Apartment density
AD NEW
Building use mix
NDVI
Apartment density
HD NEW
Building use mix
NDVI
Commercial density
AD NEW
Accessibility
Habitation density
Habitation density
Service density
service density
NDVI
commercial density
Habitation density
commercial density
Accessibility
Urban fabric
commercial density
NDVI
HD NEW
Apartment density
Accessibility accessibility
Commercial density
service density
accessibility
Habitation density
Service density
importance urban fabric
Service density
Urban fabric
Transport importance
Diversity of urban land use
The classification of urban vitality
NDVI
NDVI
In residencial, industry and transport area , urban fabric density( such as basic infrastructures ...) is the most importance to impact urban vitality for local citizens .In nature area , service density (such as local market) is the most important, and in commercial density accessibility (the distance from sites to bus/metro stations) is the most important. In a short, building a support tool is helpful to simulate stratgies for low urban vitality or empty urban spaces on different urban land use.