Multi-Criteria Group Decision Making Methods and Integrated Web-Based Decision Support Systems

Ibrahim Ozer University of Ottawa

Presentation Outline 1. Introduction

Problem

2. Methods for Multicriteria Group Decision Making 3. Decision-Making Process Diagram 4. Illustrative Problem
Best Site Selection

5. Detailed Methods 6. Web-based Tool for GDM

2

Multicriteria Group Decision Making

Introduction: Keywords

Groups make better judgments than average individual members in analysis and evaluation tasks. (McGrath, 1984; Nah & Benbasat, 1999). Never underestimate the power of stupid people in large groups. (As read on TShirts).

Multicriteria Group Decision Making

Introduction: Advantages/Disadvantages
Advantages Disadvantages

Useful in judgmental tasks Produce better decisions than individual Reducing effects of individual bias Solutions more likely to be accepted

Takes more time One capable person can decide as well as a group Satisfaction Negative effects of bias decisions Not necessarily a consensus

Problem Environmental and natural resource problems affecting coastal regions. Aspects involve marine use and ecosystem multicriteria description: Resources, Habitat, Effluents, Activities Decision Makers: Local Communities, Federal Scientists, Industrial Organizations, Non-Governmental Organizations, and Provincial Governments
Multicriteria Group Decision Making

Methods for Group Decision Making

1. 2. 3. 4. 5. 6. 7. 8.

Analytic Hierarchy Process (AHP) Weighted Sum Method (WSM) Weighted Product Method (WPM) AHP Combined Method Group Evaluation Method Fuzzy AHP Fuzzy AHP Combined Fuzzy AHP Group
Multicriteria Group Decision Making

The Decision-Making Process: Individual DMers

Identify Problem
Select Alternative

Develop Decision Criteria

Allocate Weights to Criteria

Implement Alternative

Develop Alternatives

Analyze Alternatives Evaluate Results

Multicriteria Group Decision Making

Illustrative Problem: Best Site Selection

The ABC Restaurant Corporation is offering franchise opportunities. After completing all the requirements from the applicants, the company seeks the best site location from a set of alternative locations. There are three DMs to make the judgments: CEO, CFO, and CIO.
Best Site Selection Level 0: Goal

Level 1: Criteria Visibility Accessibility Traffic Convenience

Location1 Location2 Location3

Location1 Location2 Location3

Location1 Location2 Location3

Location1 Location2 Location3

Level 2: Alternatives

Multicriteria Group Decision Making

Methods: 1. AHP
The analytic hierarchy process (AHP), which provides a proven, effective means to deal with complex decision making, was first introduced by Thomas Saaty in 1970s Evaluation phase is divided into four steps given below; 1. Generate pairwise matrices 2. Generate the weights of the measures 3. Normalize weights to get the consistency among measures 4. Calculate the overall ratings
Objective
Level 0: Goal Criteria 2 Criteria N Level 1: Criteria Alternative 1 Alternative 2 Level 2: Alternatives

Criteria 1

Multicriteria Group Decision Making

Methods: 1. AHP
Pairwise Comparison Matrix Interactive Feedback from CEO
Criteria Visibility Accessibility Traffic Convenience Visibility 1 Accessibility 3 1 Criteria ( 0.398 Visibility Loc.1 Loc.2 Loc.3 0.123 0.320 0.557 0.085 Access 0.608 0.272 0.120 0.218 Traffic 0.619 0.284 0.096 0.299 Conv. 0.265 0.656 0.080 ==> ==> ==> ) Priority 0.315 0.408 0.277 Traffic 2 1/4 1 Convenience 2 1/4 1/2 1 Priority 0.398 0.085 0.218 0.299

Multicriteria Group Decision Making

Methods: 1. AHP
Overall Location Priorities for DMs

CEO Loc.1 Loc.2 Loc.3 0.315 0.408 0.277

CFO 0.135 0.304 0.561

CIO 0.483 0.323 0.194

Multicriteria Group Decision Making

Methods: 2. WSM
Evaluation of DMs by each DM.
CEO CEO CFO CIO 1 CFO 5 1 CIO 7 3 1 Priority 0.724 0.193 0.083

(0.724
CEO Loc.1 0.315

0.193
CFO 0.135

0.083)
CIO 0.483 Priority 0.294

Loc.2
Loc.3

0.408
0.277

0.304
0.561

0.323
0.194

0.381
0.325

Overall priority of selecting a best location. Multicriteria Group Decision Making

Methods: 3. WPM
Each alternative is compared with the others by multiplying a number of ratios, one for each criterion.
Criteria ( 0.398 Visibility Loc.1 Loc.2 Loc.3 0.123 0.320 0.557 0.085 Access. 0.608 0.272 0.120 0.218 Traffic 0.619 0.284 0.096 0.299 Conv. 0.265 0.656 0.080 ==> ==> ==> R(Loc.1/Loc.2)= R(Loc.1/Loc.3)= R(Loc.2/Loc.3)= 0.660 1.349 2.043 )

Alternatives pairwise comparison matrix and priority for CEO

Since the CEO has the highest value (0.724) among the other DMs, his option -Loc.2- will be chosen to select the best location.

Multicriteria Group Decision Making

Methods: 4. AHP Combined

Geometric mean approach is used to combine the inputs of all DMs. After pairwise comparison matrix is conducted, AHP is used to get overall ranking.
Criteria Visibility Access. Traffic Conv. Priority Visibility Access.

1.00
0.38 0.72 0.72

2.62
1.00 1.75 1.44

1.39
0.57 1.00 1.10

1.39
0.69 0.91 1.00

0.357
0.149 0.247 0.247

Traffic
Convenience

Criteria pairwise comparison matrix and priority for combined. Priority Overall Loc.1 Priority Overall Loc.2 Priority Overall Loc.3 Priority 0.328 0.356 0.328

Multicriteria Group Decision Making

Methods: 5. Group Evaluation

In Group Evaluation, each DM evaluates the other DMs instead of alternatives. Each DM ranked the other two DMs with respect to criterion. Pairwise comparison matrices are created as follow
Visibility CEO CFO CIO CEO 1 CFO 5 1 CIO 8 5 1 Priority 0.711 0.223 0.066

DMs pairwise comparison matrix and priority with respect to Visibility

Multicriteria Group Decision Making

Methods: 5. Group Evaluation

Each weight of DMs is multiplied by relevant criterion to get the following pairwise comparison.

CEO

CFO

CIO

Priority 0.405

Visibility

0.398

0.503

0.145

Accessibility
Traffic

0.085
0.218

0.273
0.145

0.098
0.327

0.152
0.208

Convenience

0.299

0.079

0.430

0.314

We then weighted each alternative by multiplying their ranks by corresponding weight.

( Loc.1 Loc.2 Loc.3

0.405 0.152 0.304 0.544

0.152 0.402 0.303 0.295

0.208 0.462 0.263 0.274

0.314 0.320 0.572 0.109

) 0.319 0.403 0.356

Visibility Access. Traffic Conv.

Multicriteria Group Decision Making

Methods: 6. Fuzzy AHP

Although the AHP is to capture the experts knowledge, the traditionaly AHP still can not really reflect the human thinking style. Triangular fuzzy numbers are used based on arithmetic operations to express the decision makers evaluate on alternatives with respect to each criterion
1 1

0 a1 aM a2

0 A

Triangular Fuzzy Number and Crisp Number Multicriteria Group Decision Making

Methods: 7. Fuzzy AHP Combined

Each decision maker (DM) individually assesses alternatives and criteria following to the normal Fuzzy AHP procedures and from their assessments, the geometric mean is calculated to obtain the final decision
Criteria Visibility Access. Traffic Conv. 1 0.26 0.41 0.41 Visibility 1 0.34 0.61 0.61 1 0.53 1.05 1.05 1.93 1 2.11 1.95 Access. 2.93 1 2.72 2.62 3.93 1 3.43 3.30 1.55 0.78 1 0.76 Traffic 2.44 1.13 1 1.47 3.33 1.50 1 2.20 1.01 0.56 0.54 1 Conv. 1.71 0.74 0.81 1 2.44 0.96 1.37 1

Normalized Matrix

P1=
Visibility Access. 0.083 0.039 0.117 0.049 0.150

0.022 0.023 0.020

0.035 0.038 0.034

0.056 0.062 0.056

P2=
0.062

P3=
Traffic
Conv.

0.064
0.059

0.08
0.08

0.110
0.106

Normalized value of each criterion is multiplied by corresponding normalized alternative value and them sum them up. P2 (Loc.2) dominates the other locations.

Multicriteria Group Decision Making

Methods: 8. Fuzzy AHP Group

Weights for the DMs were empricially defined according to the AHP whereby each DM responded to the overall importance of all other DMs for this decision. Multiplying DMs judgment of criteria in fuzzy AHP by weight of each DM is called Fuzzy AHP Group.
Criteria Visibility Access. Traffic Conv. 1 0.27 0.51 0.51 Visibility 1 0.38 0.72 0.72 1 0.63 1.10 1.10 1.58 1 1.44 1.06 Access. 2.62 1 1.74 1.44 3.63 1 2.32 1.88 0.90 0.43 1 0.62 Traffic 1.38 0.57 1 1.10 1.95 0.69 1 1.65 0.90 0.53 0.60 1 Conv. 1.38 0.69 0.90 1 1.95 0.94 1.58 1

Normalized Matrix Visibility Access. Traffic Conv. 0.083 0.039 0.064 0.059 0.117 0.049 0.08 0.08 0.150 0.062 0.110 0.106

P1= P2= P3=

0.021 0.025 0.020

0.033 0.041 0.034

0.051 0.066 0.055

P2 (Loc.2) dominates the other locations.

Multicriteria Group Decision Making

Methods Strengths and Weaknesses

Methods Strength Weakness

Appropriate for GDM

Handles multiple criteria AHP Doesnt involve complex math A certain value of consistency is allowed Easy to capture and convenient WSM Strong in a single dimensional problems Eliminate any unit of measure; thus, can be used in single and multi dimensional MCDM Instead of actual values, it can use relative ones.

Perfect consistency is very difficult

Time consuming with large numbers Doesnt take into account the uncertainty

Difficulty emerges on multidimensional problems No solution with equal weight of DMs

WPM

Multicriteria Group Decision Making

Methods Strengths and Weaknesses contd

Methods Strength Weakness

AHP Combined

Simplifying the group pairwise comparisons.

Evaluating DMs

Uncertainty
Uncertainty Additive utility Time consuming Hard to convince DMs Time consuming Time consuming

Group Evaluation

Reducing the noise by having DMs weights Deals with uncertainty

Fuzzy AHP

Similar scale to Saatys can be used Reducing the # of matrices Considering the weight of DMs

Fuzzy AHP Combined Fuzzy AHP Group

Multicriteria Group Decision Making

Web-based Tool for Group Decision Making

Provincial Governments NGO
User User

WAN Database Workstation Server

User

Industrial Organizations

User User

Local Communities

Federal Scientists

Java, object-oriented programming is used.

Multicriteria Group Decision Making

Web-based Tool: Architecture of the Application

Run Time Environment, are run to evaluate the clients pairwise comparisons and then those weights are delivered to the Web browser on the client side.

Web Browser/ User Interface

Client Machine

Java Application

Run Time Environment

Windows

The weights delivered to the server and stored in the database. Java Application is used in the run-time environment to do the required calculations and results based on the appropriate methodology.

Database

Results provided using AHP Modeling

Server

Results/Reports

Multicriteria Group Decision Making

Web-based Tool: Aquawebsite

Multicriteria Group Decision Making

Web-based Tool: Aquawebsite

Multicriteria Group Decision Making

Web-based Tool: Aquawebsite

Multicriteria Group Decision Making

Thank you