Transcript
Page 1: Bolchi, Diappi, Maltese & Mariotti - input2012

ASSESSING SUSTAINABLE MOBILITY AT NEIGHBOURHOOD LEVEL Cluster Analysis and Self Organising Maps (SOM) neural network

Paola BolchiLidia DiappiIla MalteseIlaria Mariotti

DiAP, Politecnico di Milano

INPUT 2012

University of Cagliari

Cagliari, 10 - 12 / 05 / 2012

Page 2: Bolchi, Diappi, Maltese & Mariotti - input2012

STRUCTURE

• Aim of the work

• Literature review on SM and its evaluation

• Data and methodology• Data and methodology

• Descriptive statistics, Cluster Analysis and SOM Neural

Network

• Results

• Conclusion and discussion

• Further research questions

Page 3: Bolchi, Diappi, Maltese & Mariotti - input2012

AIM OF THE WORK

Investigate the SM strategies atneighbourhood level in 37European sustainableEuropean sustainableneighbourhoods.

Differences and commonalitiesamong the differentneighbourhoods, will be stressed

throughout CA and SOM NeuralNetwork.

Page 4: Bolchi, Diappi, Maltese & Mariotti - input2012

Sustainable mobility

•Allows safe basic access and development needs of

individuals, companies and society for equity within and

between successive generations (social aspects).

•Is Affordable, operates fairly and efficiently, offers a •Is Affordable, operates fairly and efficiently, offers a

choice of transport mode, and supports a competitive

economy, as well as balanced regional development

(economic aspects).

•Uses renewable resources and non-renewable

resources in a rational way, while minimizing the impact

on the use of land and the generation of noise

(environmental aspects).

(European Union Council of Ministers of Transport, 2001)

Page 5: Bolchi, Diappi, Maltese & Mariotti - input2012

Sustainable mobility: literature

1992 - 1993 1993 - 2000 2000-2005 2005-2010

Impacts on Environment Society -

quality of life

Economy -

equity

Urban

environment,

society and

economy

Disciplines Transport

economics,

transport

+ sociology + psycology,

anthropology,

political

+ planning,

urban studies,

ICT transport

geography,

environmental

engeneering

political

science

ICT

Methods Environmental

impact

assessment,

regression

analysis,

quantitative

modelling

+ scenario

building and

analysis

+ case studies,

interviews,

qualitative

modelling,

institutional

analysis

+ multi-

dimensional

(and multi-

scale)

framework ,

benchmarking

Level Macro Macro Micro/macro Micro

Question on

sustainable

mobility

What is it? When is it

sustainable?

Why is it

difficult to

achieve it?

How is it

possible to

achieve it?

How to

achieve it at

the urban and

suburban

scale?

SourceHolden (2007)

Page 6: Bolchi, Diappi, Maltese & Mariotti - input2012

SM evaluation

In the literature it is possibile to find many SM indicators at the urban scale (among the others Gilbert 2002, at the urban scale (among the others Gilbert 2002, Gundmundsson 2003, Litman 2003).

It is also easy to find indicators for the assessment of Sustainability in general, developed by international institutions (OECD, World Bank, EU, ecc...).

Here the focus is on SM indicators at a neighbourhood scale.

Page 7: Bolchi, Diappi, Maltese & Mariotti - input2012

Methodology

• 1° step – identification of SM strategies and choice of related indicators in order to create a database;database;

• 2° step – elaborating database indicators throughout Cluster Analysis and SOM Neural Network

Page 8: Bolchi, Diappi, Maltese & Mariotti - input2012

Methodology (1st step)

Nijkamp’s Hexagon (1993)

ECOWARE

Holden’s Model (2007)

ECOWARE HARDWARE

ORGWARE

SOFTWARE

CIVICWARE FINWARE

Page 9: Bolchi, Diappi, Maltese & Mariotti - input2012

The Nijkamp model

TangibleTangible

Intangible

Page 10: Bolchi, Diappi, Maltese & Mariotti - input2012

SM variables – neighbourhood scale

SM degree

ECOWARE Energy Energy saving for mobility

Transport strategies for reducing car use

Effectiveness and integration of Public Transport system

Bicycle and pedestrian paths

Efficiency of private transport system

Parking planning

Alternative fuelled vehicles

HARDWARE

Transport

Alternative fuelled vehicles

Built environement Mixed use of land

Land-use Density

Financ ing, incentives, subsidies Funds for reducing car use

Economic vitality New jobs in the mobility sector

Involvement in policies and programs for SM

Accessibility to information and inclusion in decision making processes

about SM

Partnership Public-private partnership for SM

Education and sensitizing Campaigns of communication and information about SM

Training and knowledge New sensitizing jobs (even volunteers)

Innovation Innovative approach to project and technology use for SM

CIVICWARE Partic ipation Voluntary community involvement in SM (forum, …)

FINWARE

ORGWARELocal Governance

SOFTWARE

Sources: Journals, books, magazines, Websites

Page 11: Bolchi, Diappi, Maltese & Mariotti - input2012

Direct SM indicators Indirect SM indicators

Transport strategies for reducing car use:

car sharing;

car pooling;

Funds for reducing car use

New jobs in the mobility sector

Direct and Indirect indicators

car pooling;

collective taxi;

park & ride;

bike sharing...

Involvement in policies and programs for sustainability

Accessibility to information and inclusion in

decisional making processes

Public-private partnership

Effectiveness and integration of public transport

system

Communication and information, assistance to users

Bicycle and pedestrian paths New sensitizing jobs

Private transport efficiency:

traffic calming measures

car free; ...

Innovative approach to project and technology use

Community involvement

Life quality improvement (comfort, security, air

quality, ...)

Parking planning (planning typologies: open air,

underground, ....)

Alternative fuelled vehicles

Energy saving for mobility

-road-light,

- recharging vehicles

Page 12: Bolchi, Diappi, Maltese & Mariotti - input2012

Context variables

Neighbourhood population

Neighbourhood area (kmq)

Context variables

Neighbourhood area (kmq)

Neighbourhood density

City population

City area

City density

Mixed use of land: (i) % of residential area over total

area; (ii) number of functions

Green area: % of green area over the total

GDP – NUTS3 province

Country of location

Page 13: Bolchi, Diappi, Maltese & Mariotti - input2012

Data

37 sustainableNeighbourhoods in 28 Cities in9 European Countries

• BP for sustainability• >500 inhab., >0.010 kmq

Country City

Austria Bad Ischl, Linz, Wien

Germany Freiburg, Munich, Hannover, Hamburg, Tubingen, Stuttgard

Spain Zaragoza

Finland Helsinki

Italy Torino, Roma, Modena, Reggio Emilia, • >500 inhab., >0.010 kmq• Resid. <90% tot area

Italy Torino, Roma, Modena, Reggio Emilia, Bologna, Brescia, Mantova, Bolzano,

Siena, Pesaro

The Netherlands

Amsterdam, Rotterdam

Norway Oslo

Sweden Malmo, Stockholm

UnitedKingdom

London, Perth

Page 14: Bolchi, Diappi, Maltese & Mariotti - input2012

Variables Description Measure

Characteristics of the Neighbourhood

Area Neighbourhood surface Kmq

Population Neighbourhood inhabitants Number of inhabitants

Density Population / surface n./kmq

North Europe If the neigbourhood is located in Northen Europe Dummy variable: 0, 1

Residential

area

Share of residential surface over totalsurface %

Mix Number of functions present in the neighbourhood 1 to 6

Green area Share of green area over the total surface %

SM indicators at neighbourhood level

Energy saving Energy saving for mobility 1 to 3

Transp.

Reduct.

Transport strategies for reducing car use 1 to 3

Lpt Effectiveness and integration of public transport system 1 to 3 Bicycle paths Bicycle and pedestrian paths 1 to 3 Efficient

Planning

Private transport efficiency 1 to 3

Parking Parking planning 1 to 3

SourceUrban AuditEurostat

Parking

Planning

Parking planning 1 to 3

Alternative

fuelled vehicles

Alternative fuelled vehicles 1 to 3

SM average Average value of the SM indicators, excluding access to information,

sensitivity and community involvement

1 to 3

Access to

information

Accessibility to information and inclusion in decision making processes 1 to 3

Sensitizing Communication and information, assistance to users 1 to 3 Involvment Community involvement 1 to 3 Sens_Inv Communication and information, assistance to users and community

involvement (average) 1 to 3

Indicators at urban level

Area City area Kmq

Population City inhabitants Number of inhabitants

Density City Population / area n./kmq

Indicator at NUTS 3 province

GDP 1998 GDP at the year 1998 Euros / Source: Eurostat

Sources: Journals, books, magazines, Websites

Page 15: Bolchi, Diappi, Maltese & Mariotti - input2012

Methodology (2nd step)

• CLUSTER ANALYSIS – based on linear models,

WELL COMPARED TOWELL COMPARED TO

• SELF ORGANISING MAPS (SOM) neural network –adaptive non-linear method

Page 16: Bolchi, Diappi, Maltese & Mariotti - input2012
Page 17: Bolchi, Diappi, Maltese & Mariotti - input2012

CA results: neighbourhoods.

5a) Neighbourhood

Cluster Area Pop. Density Resid. Mix

SM

average Access.

Sens

Inv

North

Europe

Green

area

1 .28 2703.85 14444.01 .73 2.90 2.32 2.61 2.57 .38 .33 1 .28 2703.85 14444.01 .73 2.90 2.32 2.61 2.57 .38 .33

2 1.9 10087.56 20367.39 .63 3.77 2.36 2.44 2.44 .78 .34

3 1.4 5082 20645.56 .77 2.33 2.26 2.66 2.33 .67 .28

4 .24 8850 36875 .5 6 2.42 2 2 1 .24

Media .86 5051.64 17496.74 .70 3.10 2.32 2.56 2.48 .54 .33

Page 18: Bolchi, Diappi, Maltese & Mariotti - input2012

CA results cities and NUTS3

5b) City NUTS 3

Cluster Area Pop. Density GDP_1998

1 137.71 152150.7 1294.15 33223.81

2 455.44 643759.3 2870.41 40791.22

3 800.83 1900245 2738.95 49200 3 800.83 1900245 2738.95 49200

4 8760 7413100 846.24 50600

Media

555.57

751447.8

1899.75

38124.89

Page 19: Bolchi, Diappi, Maltese & Mariotti - input2012

CA results SM

5c) Neigbourhood – SM indicators

Direct SM indicators

Indirect SM indicators

Cluster

Transp.

Reduct.

Lpt Bicycle

paths

Efficient

Planning

Parking

Planning

Alternative

fuelled

vehicles

Energy

saving

Access to

information

Sensitivity Community

involvment

1 2.47 2.62 3 2.24 2.28 1.86 1.80 2.62 2.38 2.52 1 2.47 2.62 3 2.24 2.28 1.86 1.80 2.62 2.38 2.52

2 2.33 2.55 2.89 2.22 2.22 2.22 2.11 2.44 2.44 2.66

3 2.33 2.83 2.83 2.33 2.33 1.66 1.5 2.66 2.33 1.83

4 3 3 3 2 3 1 2 2 2 3

Media 2.43 2.65 2.94 2.24 2.30 1.90 1.84 2.57 2.38 2.46

Page 20: Bolchi, Diappi, Maltese & Mariotti - input2012

CA results

CL SM ACCESS Neigh.

SIZE

MIX density GREEN CITY

SIZE

GDP Eu POS

1 + ++ -- -- -- -- -- -- South

2 ++ + ++ + + ++ + -- Central-

North North

3 - ++ ++ -- + -- ++ ++ Both

4 ++ -- -- ++ ++ -- ++ ++ North

Page 21: Bolchi, Diappi, Maltese & Mariotti - input2012

CA results

Cluster Description Neighbourhoods Cluster 1 Medium values for SM; high access to

information and inclusion in decisional making

processes, sensitizing to sustainable mobility and community involvement. Small neighbourhoods, low dense. Residential area prevails and very

little different functions. Small green area. Small cities and low dense; low GDP, compared to the

average, mainly located in the South of Europe.

Amschl, Bo01, Borgo delle corti, Casanova, Cognento, Ecocity Bad

ischl, Ecocity Umbertide, Ecocity Tubingen, Fairfield, Giuncoli, Lunetta, Malizia, Parco Ottavi,

Rieselfeld, S.Francesco Biopep, San Rocco, San Pietro, Solar city,

Vauban, Villa Fastiggi, Violino

Cluster 2 Good values for SM, medium values for accessibility to information, sensitizing to sustainable mobility and community

Burgholzhof, Gwl, Hammarby, Kronsberg, Nieuw Terbregge, Pilastredet, Valdespartera, Vikki, sustainable mobility and community

involvement. Large neighbourhoods with medium density values; large green area and

number of func tions on average. Medium-small cities, with lower GDP, mainly located in the

centre-north of Europe.

Pilastredet, Valdespartera, Vikki,

Villaggio Olimpico

Cluster 3 Medium-low SM values (the lowest). High value for accessibility to information but low va lue for sensitising and community involvment. Large

neighbourhoods with medium density. Residential area prevails, and little number of

functions. Small surface of green areas. Cities medium-large with high density and high GDP, located both in the centre-north and in the south.

Falkenried-Terrassen, Hafencity, Lunghezzina, Nordmanngasse, Parco Plinio, Riem

Cluster 4 Good values for SM, low values for accessibility

and sensitizing and community involvement. Very dense neighbourhoods (small in surface)

and mixed used. Little percentage of green area. Large cities low dense but with the highest GDP on average. All located in the north.

Gmv

Page 22: Bolchi, Diappi, Maltese & Mariotti - input2012

Neural NetworkNeural Network

Page 23: Bolchi, Diappi, Maltese & Mariotti - input2012

The functioning of the SOM RN: The network is deformed by the learning algorithm to

bring the nodes close to the groups of observations

a ba b

Page 24: Bolchi, Diappi, Maltese & Mariotti - input2012

c11

c12

0.0

0.2

0.4

0.6

0.8

1.0

c13

0.0

0.2

0.4

0.6

0.8

1.0

c21 c22

0.2

0.4

0.6

0.8

1.0

c23

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.0

0.2

c31

c32

0.0

0.2

0.4

0.6

0.8

1.0

c33

0.0

0.2

0.4

0.6

0.8

1.0

Page 25: Bolchi, Diappi, Maltese & Mariotti - input2012
Page 26: Bolchi, Diappi, Maltese & Mariotti - input2012

SOM 31

c31

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

supQ

km

q

popQ

NEW

DensQ

Verd

ePerc

Resid

mix

NEW

en_sav

tranre

dcar

tpl

bic

yc

effic

ptran

park

p

alte

rfv

access

sens

involv

km

q c

ity

pop_city

Dcity

GD

P_1998

NorthEU

supQ

km

q

NEW

DensQ

Verd

ePerc

mix

NEW

en_sav

tranre

dcar

effic

ptran

access

km

q c

ity

pop_city

GD

P_1998

NorthEU

MIX SM ACC N.size C.size GDP Pos

CL2++ ++ ++ ++ -- -- NorthSOM31

Page 27: Bolchi, Diappi, Maltese & Mariotti - input2012

SOM 23c23

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

supQ

km

q

popQ

NEW

DensQ

Verd

ePerc

Resid

mix

NEW

en_sav

tranre

dcar

tpl

bic

yc

effic

ptran

park

p

alte

rfv

access

sens

involv

km

q c

ity

pop_city

Dcity

GD

P_1998

NorthEU

supQ

km

q

popQ

NEW

DensQ

Verd

ePerc

Resid

mix

NEW

en_sav

tranre

dcar

bic

yc

effic

ptran

park

p

alte

rfv

access

sens

involv

km

q c

ity

pop_city

Dcity

GD

P_1998

NorthEU

MIX SM ACC N.size C.size GDP Pos

CL1-- + ++ -- -- -- SouthSOM23

Page 28: Bolchi, Diappi, Maltese & Mariotti - input2012

CONCLUSIONS

• Best north-european performance;

• Best “sustainable mobility” practices are those neighbourhoods which invested a lot in an omogeneous way on all the indicators, both direct and indirect;and indirect;

• Mixitè appears more significant than density and also the presence of green areas.

• In general citizens’ participation is fundamental

• New technologies don’t appear as the most adopted tool for achieving sustainable mobility: land use and green attitudes are preferred;

• Context variables don’t explain so much

• Two methods quite “agree”, despite some differences in selecting elements and grouping them

Page 29: Bolchi, Diappi, Maltese & Mariotti - input2012

FURTHER RESEARCH

• Further analysis of FINWARE: income and incentives• GDP: it appears useful to be better analysed• Direct vs indirect also is an interesting topic• Direct vs indirect also is an interesting topic• Quality and type of the variables (discrete and continuous)• Further analysis of other context characteristics: presence of infrastructures• Freight transport could worth be analysed, because it is a key factor for achieving a real and complete sustainable mobility: further analysis on city logistics• Some case study with analysis of citizen’s satisfaction

Page 30: Bolchi, Diappi, Maltese & Mariotti - input2012

Suggestions are welcome!

Paola BolchiLidia DiappiIla MalteseIlaria Mariotti

DiAP, Politecnico di Milano

[email protected]

Page 31: Bolchi, Diappi, Maltese & Mariotti - input2012

San Pietro Bologna IT

Casanova Bolzano IT

Violino Brescia IT

San Rocco Faenza IT

Giuncoli Firenze IT

Amschl Freiburg DE

Vauban Freiburg DE

Rieselfeld Freiburg DE

Falkenried-Terrassen Hamburg DE

Hafencity Hamburg DE

Kronsberg Hannover DE

Vikki Helsinki FI

Solar c ity Linz AU

Gmv London UK

Bo01 Malmo SW

Lunetta Mantova IT

S.Francesco Biopep Nonatola - MO IT

Cognento Modena IT Cognento Modena IT

Borgo delle corti Modena IT

Riem Monaco DE

Pilestredet Oslo NW

Fairfield Perth UK

Villa Fastiggi Pesaro IT

Parco Ottavi Reggio E. IT

Parco Plinio Roma IT

Lunghezzina Roma IT

Nieuw Terbregge Rotterdam NL

Malizia Siena IT

Hammarby Stocholm SW

Burgholzhof Stuttga rd DE

Villaggio olimpico Torino IT

Ecocity Tubingen Tubingen DE

Ecocity Umbertide Umbertide IT

Nordmanngasse Wien AU

Valdespartera Zaragoza ES


Top Related