de wijkwizard

Léon van Berlo / Jos van Leeuwen

The Neighbourhood Wizard

Cause and effect of changes in urban neighbourhoods

05 July 2006 2

Agenda

• Introduction• Objective• Approach• (Experiencing) Liveability• Data Collection• Knowledge representation• Prototype• Evaluation and testing• Conclusions and future work• Coffee break

05 July 2006 3

Introduction

• Quality of the neighbourhood (physical and social) Increasingly important

• Local initiatives for neighbourhood improvement

• Municipalities support these initiatives Citizen participation

• Issues: • Inhabitants focus on their own problems (not the ones from

their neighbours)• Inhabitants don’t see the complex dependencies of a decision• Inhabitants give concrete proposals for change in stead of

their desire

05 July 2006 4

Objective

• Making citizens realise what the consequences are of their ideas for changes

• By developing a tool that allows citizens to:- propose changes to their neighbourhood;- assess the quality of these changes

05 July 2006 5

Approach

• Find a set of indicators for experience of liveability of the neighbourhood

• Find a set of characteristics that affect the liveability

• Determine a BBN that represents the knowledge

• Build a prototype • Narrowing its scope to the plaza type of habitat

• Testing the prototype in the Dutch city of ’s Hertogenbosch‑

05 July 2006 6

Experiencing liveability:Leidelmeijer and Marsman 1999

Environment characteristic 1

Environment characteristic ..

Environment characteristic n

Appreciation 1

Appreciation n

Importance ..

Appreciation ..

Importance 1

Importance n

Satisfaction 1

Satisfaction ..

Satisfaction n

Experienced liveability

x

x

x

x

x

x

+

+

+

Infulenced by the habitat

OBJECTIVE SUBJECTIVE

05 July 2006 7

Example experience by an individual

Composition:simple

Status:popular

Security:safe

neutral

positive

none

negative

low+/-

++

++

xx

xx

xx

+

+

+

high

AppreciationImportance

Satisfaction LiveabilityCharacteristics

05 July 2006 8


Composition:simple

Status:popular

Security:safe

neutral

positive

none

negative

low+/-

++

Exp.liveability

x

x

x

x

x

x+

+

+

high

Charcteristic ..importance

appreciation

xx

+


appreciation

--x

x+


appreciation

xx

+


appreciation

xx

+

05 July 2006 9


ASPECT

ELEMENT

ASPECT

neutral

positive

low+/-

++

Exp.liveability

x

x

x

x+

+

high

ELEMENT

ASPECTimportance

appreciation

--x

x+

ELEMENT

ELEMENT

05 July 2006 10

Experience by another individual

ASPECT

ELEMENT

ASPECT

neutral

positive

low+/-

+

Exp.liveability

xx

x

x+

+

normal

ELEMENT

ASPECTimportance

appreciation

--xx

+

ELEMENT

ELEMENT

05 July 2006 11

Grouping individuals and their needs

Wishprofiles:• Teenagers• Yuppies• Families• Elderly• Handicapped (elderly)

Aspects:• Space• Liveliness• Security• Quality• Status• Traffic

05 July 2006 12

Data Collection

• Questionnaire of liveability regarding the city of ’s Hertogenbosch ‑• Experiences of characteristics such as:

• ‘public furnishing’• ‘available facilities’• ‘public accessibility’• ‘status’• ‘appearance’• ‘ambiance’• etc.

• For plazas, over 40 characteristics were included. • Scale of seven possible values

• Ranging from deficient, through moderate and neutral, to ample and excessive.

05 July 2006 13

Data Collection

Example:

Form and function:IncoherentSuitablesurprising conflicting

05 July 2006 14

Knowledge Representation

• Bayesian Network: • Can deal with uncertainty and interdependent variables

• Determining the structure of a BN:• 1) Knowledge expert who constructs a network • 2) Examining data from the particular domain

• In this project 2 is used to come to a base network which was refined by 1.

05 July 2006 15

Structural Learning

• Hugin (www.hugin.com) was used with:• PC algorithm (Peter & Clark)

• NPC algorithm (Necessary Path Condition)

• Constraint-based learning algorithms

• Derive conditional independence and dependence statements by performing statistical tests on pairs of variables in the data set

05 July 2006 16

BN Structure (1)

05 July 2006 17

Structural Learning

• PC and NPC same results• Significance level 0.05 – 0.03 – 0.01

• Difference in ‘real relationships’ and ‘relationships in the data’

• Defining relations that are not in the data: no use

05 July 2006 18

BN Structure (2)

05 July 2006 19

Prototype

• User-interaction focused on a task assigned to the user• Users can experience this like a game

• Representing the effects of changes

• Representing the desired states of the aspects for different sections of the population

• Availability of the system on Internet

• Easy to use interface and obvious navigation

05 July 2006 20

Changing elements:

• Three ways:• 1) Drawing• 2) Picking from a list• 3) Cheating

05 July 2006 21

Changing elements: 1 (drawing)

05 July 2006 22

Changing elements: 2 (picking from a list)

05 July 2006 23

Changing elements: 3 (cheating)

05 July 2006 24

Presentation of Predicted Effects

• Three levels:• 1) Simple does not give desired effect• 2) Normal• 3) Expert

05 July 2006 25

Presentation of Predicted Effects: 2 (normal)

05 July 2006 26

Presentation of Predicted Effects: 2 (normal)

05 July 2006 27

Presentation of Predicted Effects: 3 (expert)

05 July 2006 28

Presentation of Predicted Effects: 3 (expert)

05 July 2006 29

Evaluation

• www.WijkWizard.nl (dutch)

• Tested and evaluated by inhabitants of the city of ’s Hertogenbosch. ‑• Online evaluation form.

• “Thanks to the Neighbourhood Wizard, I now see that certain ideas are positive for me, but negative for other members of our community” : 7.4

• “The Neighbourhood Wizard shows me that changes can have positive effects on one aspect, but negative effects on other aspects” : 7.0

• Confirmed the educational function of the prototype!

05 July 2006 30

Conclusions (+)

• The Neighbourhood Wizard helps users to see that certain ideas are positive for them, but negative for other sections of the population;

• The Neighbourhood Wizard shows users that changes can have positive effects on one aspect, but negative effects on other aspects;

• The Neighbourhood Wizard helps users to realize the complexity of a design task and as a result users will have a better informed view on plan proposals and probably a higher appreciation of plans.

05 July 2006 31

Conclusions (-)

• Design of the user interface

• Navigation structure (too many clicks)

• Abstract terms• Inclusion of more concrete elements (number of parking lots)

can help take away long-living irritations that inhabitants may have

• The data collection is restricted to physical characteristics

05 July 2006 32

Future work

• Investigate the relations between characteristics in depth

• (developing a technique that) Includes explanations of the effects• In some cases the predictions are not so obvious and require

further explanation

For example: The creation of a quiet plaza has negative effects on the safety of the plaza. This is not a logical, though correct, prediction because the quietness of a plaza will attract criminal behaviour

05 July 2006 33

Thank you

• Questions or coffee break?

[email protected] / WijkWizard.nl [email protected] / www.ddss.nl

de wijkwizard

Technology