Download presentation
Presentation is loading. Please wait.
1
SUPPORTING BUSINESS DECISIONS A Case Study: Multi-Criteria Decision Modelling
2
CASE: The Macrosoft Conference Cathy Smith, Promotions Manager of Macrosoft Software, must find a venue for this year’s Macrosoft User Group conference Already agreed: It must take place in mid-December in or around Leeds
3
Problem Definition Scope: –C–Choose a conference venue in the Leeds area Options: –A–All centres within 10 miles of City Square capable of accommodating 100 delegates Criteria: –L–Location –F–Facilities –C–Cost
4
A pedantic aside The singular is “criterion” The plural is “criteria” Hence we have one criterion, two or more criteria Please don’t write “criterions”!
5
A Hierarchy of Criteria Level 1 –LOCATION –FACILITIES –COST Level 2 –Environment –Access –Staff experience –Conference rooms –Social meeting places –Accommodation –Catering
6
The Options A ‘quick and dirty’ application of the criteria has resulted in the following shortlist: A: The Metropole Hotel B: The Chevin Lodge C: Leeds Met’s Headingley Campus D: Cookridge Hall Leisure Centre
7
AB CD
8
Importance Weights …indicate the relative importance of the criteria Several sets of weights are needed: one for the Level 1 criteria one for the set of Level 2 criteria associated with each Level 1 criterion Weights are based on ‘best judgement’ –J–Judgement always(?) needed in decision-making –I–Ideally Cathy will consult others in arriving at the judgements Each set must be normalised, i.e. add up to 1 This ensures that the criteria contribute appropriately to the assessment of the options
9
Achievement Scores …indicate how well each option (ie venue) ‘performs’ relative to the others against each Level 2 criterion Scores are based on ‘best available opinion’ –J–Judgement comes in again therefore Against a given criterion: the best option scores 100 (there could be more than one with a score of 100) the worst option scores 0 other scores fall between 0 and 100
10
Evaluating the Options Level 2 importance weights and achievement scores are now used to compute: a Level 1 criterion score for each option These, together with Level 1 importance weights, are then used to compute: an overall score for each option The highest aggregate score indicates the preferred option –N–Note that it has taken account of all the criteria
11
A Simple DSS …helps to solve Cathy’s problem It provides: a clear display of criteria, options, weights and scores the computations needed for Level 1 scores and overall scores graphical displays if required opportunity for further analysis It’s available on X-stream and H drive
12
CONFERENCE VENUE SELECTION CRITERIA WEIGHTS VENUE SCORES Level 1Level 2ABCD Location 0.35 Environment 0.5002540100 Access 0.5010080350 Facilities 0.45 Staff experience 0.1010080400 Conference rooms 0.3075100500 Social meeting places 0.1525100075 Accommodation 0.2080100090 Catering 0.2560700100 Cost 0.20 Amounts (£)9,75010,0005,0007,000 Scores5010060 EVALUATION Location50.052.537.550.0 Facilities67.390.519.054.3 Cost 5.00.0100.060.0 Overall 48.859.141.753.9 Rank 3142
13
Subjectivity in MCDM The role of subjective judgement is highlighted by MCDM in the selection of: the criteria themselves importance weights achievement scores Such judgements are often unreliable, even with expert advice To counter this, decision makers often use…
14
Sensitivity Analysis …is a way of testing the robustness of the preferred option against changes in the data / judgements on which the decision is made In MCDM, it is wise to test the decision against changes in importance weights and achievement scores This may: increase confidence in the preferred option If there is little change indicate areas in which further thought/data are needed If a small change to an input leads to large change in overall scores
15
Assignment FAQ (1) Is the assignment criteria referenced or norm referenced? –C–Criteria referenced –S–So – read the criteria!
16
FAQ (2) Does the MCDM model need to “work”? –For the better marks, yes. –A pass mark for this part can be gained by a well-designed front end only –The more sophisticated the model, the better the mark is likely to be –Please remember to submit the model on CD
17
FAQ (3 and 4) Are the word limits sacrosanct? –Y–Yes, but can go 10% above. –(–(There is no lower limit!) When should I start? –Y–You should have already started
18
FAQ (5) How do I do this part: “justify the choice of decision criteria in your model”? –Y–You need to do some research on the sort of factors that might be taken into consideration. –R–Remember to give appropriate citations (Harvard style)
19
FAQ (6) How do I do this part: –“–“Critically assess each of these technologies in relation to Bill’s current decision: Expert systems Neural networks”? Research again needed! (Can use lecture material of course).
20
FAQ (6) continued One possible approach: –D–Define the technology –L–Look at problem characteristics discussed in the relevant lecture –L–Look at “advantages” and “disadvantages” discussed in the relevant lecture –L–Look for any precedents (research!)
21
FAQ (7 and 8) Should the rules show how the amount of benefit is worked out? –Y–Yes Should the and/or tree show how the amount of benefit is worked out? –I–In cut-down mode, yes. –O–One approach – make the rightmost branch “calculations done” –T–There is a more elegant approach – worth 3 marks!
22
Nota Bene The deadline is 1500 on Friday October 24 2008. However, this is not really feasible, given the lecture schedule. Hence, there is a new deadline: –1–1500 on Friday November 7, 2008
23
Please also note… You can still hand in on 1500 on Friday October 24 2008 if you want! It remains the case that: –T–Tempus fugit
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.