Introduction Methodology Implementation Inference Demonstration Conclusion Integration of decision aid tools in a Geographical Information System Olivier Sobrie University of Mons Faculty of engineering June 22, 2011
Introduction Methodology Implementation Inference Demonstration Conclusion Introduction 1 Methodology 2 Implementation 3 Inference 4 Demonstration 5 Conclusion 6
Introduction Methodology Implementation Inference Demonstration Conclusion GIS and MCDA Combination Spatial Query GIS Visualization Organization Prediction Analysis ◮ GIS are used in lot of application from land suitability problem to geomarketing ◮ Since 90’s, works about GIS and MCDA ◮ Not a lot of work based on ELECTRE methods ◮ ELECTRE methods fit well for ordinal problems
Introduction Methodology Implementation Inference Demonstration Conclusion GIS and MCDA Limitations of GIS-MCDA works according to S. Chakhar : ◮ Weak coupling ◮ One MCDA method integrated (Single criterion synthesis) ◮ Choice of the MCDA method ◮ User’s knowledge of GIS and MCDA
Introduction Methodology Implementation Inference Demonstration Conclusion GIS and MCDA Limitations of GIS-MCDA works according to S. Chakhar : ◮ Weak coupling ◮ One MCDA method integrated (Single criterion synthesis) ◮ Choice of the MCDA method ◮ User’s knowledge of GIS and MCDA We add an extra one : A good number of GIS-MCDA tools were abandoned or never surpassed the stage of prototype. Moreover it has been done in commercial GIS.
Introduction Methodology Implementation Inference Demonstration Conclusion Objectives of our GIS-MCDA integration First objectives ◮ ELECTRE TRI implementation ◮ Tight coupling ◮ User friendly interface ◮ Open Source GIS (and implementation)
Introduction Methodology Implementation Inference Demonstration Conclusion Objectives of our GIS-MCDA integration First objectives ◮ ELECTRE TRI implementation ◮ Tight coupling ◮ User friendly interface ◮ Open Source GIS (and implementation) Second objectives ◮ Learning of parameters ◮ Implementation of a XMCDA webservice ◮ Experimentations ◮ Coupling with the ELECTRE TRI plugin
Introduction Methodology Implementation Inference Demonstration Conclusion ELECTRE TRI Parameters ◮ weights C p − 1 C p C 1 C 2 g n ◮ profiles g n − 1 ◮ credibility threshold g n − 2 ◮ ... g 2 Approach ◮ Classical g 1 b 1 b 2 b p − 2 b p − 1 b 0 b p ◮ Bouyssou-Marchant Major interests ◮ Judge an action independently from the others ◮ Allow to consider more actions than other ELECTRE methods ◮ Reference values fixed : profiles
Introduction Methodology Implementation Inference Demonstration Conclusion Application : Densification of Quebec city Subject Quebec city wants to create a program to densify its population in the centrum and around the small crown. The program consists to build rental properties at low prices for young families in empty areas. Objectives ◮ Densify central sectors where there are more public transports ◮ Sustain a good social diversity by choosing in priority the sectors where young people and immigrants are not well represented ◮ Favor sectors with a lot of small shops
Introduction Methodology Implementation Inference Demonstration Conclusion Application : Densification of Quebec city Decision map
Introduction Methodology Implementation Inference Demonstration Conclusion Application : Densification of Quebec city Definition of the problem Actions 786 districts (polygons) Criteria ◮ Density of 0-14 years old [%] (min) ◮ Density of shops [shops/ha] (max) ◮ Density of people [residents/ha] (min) ◮ Level of public transports (average) [bus/hour] (max) ◮ Ratio of immigrants [%] (min)
Introduction Methodology Implementation Inference Demonstration Conclusion Application : Densification of Quebec city Performance table
Introduction Methodology Implementation Inference Demonstration Conclusion Strategy of integration Reference ◮ Chakhar’s thesis (2006) Coupling strategy ◮ Malczewski (2006) reports only 10 % of works using a strategy of tight coupling of the MCDA method in the GIS ◮ Tight coupling Actions and criteria ◮ Vector layer ◮ actions = points, lines, polygons ◮ criteria = attributes
Introduction Methodology Implementation Inference Demonstration Conclusion Strategy to build the decision map Criterion map 1 Criterion map 2 Criterion map 3 Step 1: Construction of criterion maps Multicriteria map Step 2: Construction of the multicriteria map ELECTRE TRI Inference Step 3: ELECTRE TRI model module module Decision map Step 4: Generation of the decision map
Introduction Methodology Implementation Inference Demonstration Conclusion Choice of the GIS Requirements ◮ Open Source GIS and implementation ◮ User friendly interface ◮ Support of vector layer ◮ With map algebra tools
Introduction Methodology Implementation Inference Demonstration Conclusion Choice of the GIS Requirements ◮ Open Source GIS and implementation ◮ User friendly interface ◮ Support of vector layer ◮ With map algebra tools Lot of open source GIS ◮ GRASS, PostGIS, Quantum GIS ◮ http://opensourcegis.org/
Introduction Methodology Implementation Inference Demonstration Conclusion Quantum GIS Characteristics ◮ Great portability (Linux, Windows, Mac OS) ◮ Plugin mechanism ◮ Lot of functionnalities (GRASS, map algebra, ...) ◮ User-friendly interface
Introduction Methodology Implementation Inference Demonstration Conclusion ELECTRE TRI plugin Tight coupling
Introduction Methodology Implementation Inference Demonstration Conclusion ELECTRE TRI plugin User interface
Introduction Methodology Implementation Inference Demonstration Conclusion ELECTRE TRI plugin User interface
Introduction Methodology Implementation Inference Demonstration Conclusion ELECTRE TRI plugin User interface
Introduction Methodology Implementation Inference Demonstration Conclusion XMCDA webservice Categories profiles Learning alternatives Performance table of profiles Criteria Criteria weights XMCDA Performance table webservice Credibility threshold Categories Compatible alternatives Affectations Message Characteristics ◮ Based on A. Leroy master thesis (2010) ◮ Learning of ELECTRE TRI Bouyssou-Marchant parameters ◮ Accept non-admissible set of learning alternatives ◮ Maximize number of compatible alternatives ◮ MIP problem ◮ Use GLPK
Introduction Methodology Implementation Inference Demonstration Conclusion ELECTRE TRI BM inference experimentations First conclusions ◮ Lot of learning alternatives needed to get good results ◮ Difficult to get good set of params when learning set not completely compatible with ELECTRE TRI model ◮ Computing time becomes huge when number of learning alternatives increases
Introduction Methodology Implementation Inference Demonstration Conclusion ELECTRE TRI BM inference experimentations First conclusions ◮ Lot of learning alternatives needed to get good results ◮ Difficult to get good set of params when learning set not completely compatible with ELECTRE TRI model ◮ Computing time becomes huge when number of learning alternatives increases New experimentations ◮ Two step inference ◮ Partial inference ◮ Improve objective of the inference program
Introduction Methodology Implementation Inference Demonstration Conclusion ELECTRE TRI BM inference webservice update Learning alternatives Criteria Categories profiles Performances table Performance table of profiles Categories Criteria weights XMCDA Affectations webservice Credibility threshold Categories profiles ( a ) Compatible alternatives Performance table of profiles Message Criteria weights ( b ) Credibility threshold Characteristics ◮ Two entries added to do partial inference of the weights and lambda threshold ◮ Two entries added to do partial inference of the profiles
Introduction Methodology Implementation Inference Demonstration Conclusion Webservice available in diviz
Introduction Methodology Implementation Inference Demonstration Conclusion Coupling of XMCDA webservice with Quantum GIS ELECTRE TRI plugin Quantum GIS Main functionnal- Solver ities of the GIS ELECTRE TRI XMCDA messages XMCDA plugin webservice XMCDA files
Introduction Methodology Implementation Inference Demonstration Conclusion It’s time for the demo...
Introduction Methodology Implementation Inference Demonstration Conclusion Original model
Introduction Methodology Implementation Inference Demonstration Conclusion Actions of reference
Introduction Methodology Implementation Inference Demonstration Conclusion Global inference
Introduction Methodology Implementation Inference Demonstration Conclusion Global inference (difference) ± 29% of invalid affectations
Introduction Methodology Implementation Inference Demonstration Conclusion Profiles inference
Introduction Methodology Implementation Inference Demonstration Conclusion Profiles inference (difference) ± 33% of invalid affectations
Introduction Methodology Implementation Inference Demonstration Conclusion Weights and lambda inference
Recommend
More recommend