Professional Documents
Culture Documents
_______________________________________________________
As it is aptly called, the bSURE Strategic Decision Support Tool is a computerized information
system that combines data, analytical tools and models to support semi-structured and
unstructured management decision-making.
It can be used in the following stages of decision-making process.
o Identify and define potential problems and/or opportunities,
o Develop alternative solutions to the problem, and
o Select an alternative or a course of action.
As it is a decision support for a rationale decision maker, bSURE decision support tool or system
offers a substantial amount of analytical tools for user to evaluate the robustness of a decision
model.
2 Why AHP?
When evaluating complex issues, factors contributing to the issue are not always equally
contributing to the objective. Many a time, decision makers tend to make all factors equally
contributable making the final judgment skewed.
We use a known technique devised by Professor Thomas Saaty called Analytic Hierarchy
Process (AHP), which simplifies the calculation of weights for each action to the mission or
objective (Saaty, 2000 and Islam, 2003).
As these weights are ratio scales, they can therefore be used to leverage the weight of
contribution to the cost to the benefits in a BOCR (Benefits-Opportunities-Costs-Risks) decision
model, making decision making more systematic and justifiable (Saaty, 2005).
Making consensus decisions are now more systematic, objective, transparent, efficient and
effective.
For more details on this process, there are many websites that contain the relevant information.
The main window can be resized according to individual user’s needs. Click the maximize button
The user has the option to use the top toolbar or from the floating menu for the same features.
Click “Disable Toolbar” or “Enable Toolbar “to hide or show the toolbar.
Use the following buttons for specific operations:
New Create new issue
Open existing file from
Open
directory
SaveAs Save current file
In order to continue this user training better, let’s assume a problem. Let’s assume we want to
buy a car. Let’s assume a decision model that is based on the following hierarchy:
Mission
Choose a Car
As you notice, in this presentation, same list of Alternatives is repeated for all criteria, meaning no
scales are required for this model. The relative weights for each criterion are calculated with
respect to the criterion from relative comparison of each alternative against another on pairwise
or set of two-basis.
This is good because hard data are sometimes difficult to come by but we need to rely on
subjective parameters. Humans have a good sense of notion and can judge if a car is more
comfortable than the other by just taking a feel at a show room.
We could calculate the weights by substituting the hard data by comparing instead the
alternatives in respect to the criterion. This require us using the AHP technique.
right of Goal :
Type “Buy Best Car” and click the tick button to save.
4.3 Treeview
The Treeview is where we insert the criterion.
For a floating menu, you can right-hand click the root node (at the ‘Goal’ text) or at any nodes.
Select Add Node, to add the first criteria. Type “Cost” and click OK.
Type in the description of the criterion. These details will be displayed as Tool tip.
Continue with the other criteria and click cancel after the last criterion has been entered. The
Treeview will now look like this:
You may want to save your work at this stage. For the sake of completeness, name the file
“BuyCarExample.bsg”.
The nodes are not checked if they are disabled. We can now proceed to calculate the weights
using the AHP. First, we must ensure the enable the criteria are enabled before proceeding to the
next stage. To enable, click the relevant checkboxes as follow:
Select the level of importance of one criterion compare to the other based on the following scale:
Use button to invert the comparison. Inverting the comparison will change the way the
questionnaire is read, i.e.:
Factor B, instead, is more important:
Use the following buttons for the specific operations of the form:
Next Comparison
Once either End button is clicked, the comparison process has basically ended, and the user is
then prompted with the following Analyze Pairwise Results form:
Clicking the Skip button will skip the comparison process and bring us to the same form.
The blue progress bars on the right display the corresponding weights:
The weights are displayed when the mouse hovers on top of the corresponding bar. The cells
represent the following Pairwise Comparison Matrix and should be equal to the levels selected
during the pairwise comparison process.
COS Rel Com Saf
COST 1 1/4 3 1
Reliability 4 1 2 2
Safety 1 1/2 1 1
Hover the mouse on top of the column title (in this example, COS), for immediate details (tooltip)
of the criteria, on the progress bar for the detailed weights of the criteria.
If you notice, the consistency index (C.I) is coloured red since the value is greater than 0.1. For a
good comparison, the C.I should be less than 0.1. The software has a function that will advise the
user the specific cell that will give the most gain in terms of consistency when the value is
changed.
The values of the cells are not always based on integers (i.e. 1, 2, 3, etc) as you would see later.
These values can still be changed by double clicking the specific cell. Only the diagonal cells,
which are always 1, are not modifiable.
Use the following command or menu buttons for the specific operations:
To improve the consistency, recalculate the current matrix, by pressing . Nevertheless, the user
may want to proceed with the current values; only that the decision making process will be
consistently inconsistent; so to speak.
The textbox at the bottom of the Matrix keeps the historical information of previous comparisons.
Once ‘Save’ is initiated, the weights are displayed at the main form as the cell Local Weights
(LW). The system will automatically calculate the Global Weights (GW):
The above implements the same following Pairwise Comparison Matrix. Take special note of the
fractional values and its position in the above input format.
COS Rel Com Saf
COST 1 1/4 3 1
Reliability 4 1 2 2
Safety 1 1/2 1 1
Select “Add Option” to add the first alternative, “WAJA”. Continuously, add the rest and
terminate the list by selecting “Cancel” after the last alternative.
There are 2 ways of doing the comparison; by relative pairwise comparison or by relative data
comparison. Suppose that we have the Sticker Prices from the ever eager salesmen, we could
directly put these values into the system.
From the left Criteria List Box, select “COST” and then from the form menu, select Relative-
>Data:
You can select either normal or inverse mode. For this example, as it is cost
factor, where smaller value is better, select from the form menu, Evaluate-
>Inverse; and click OK to proceed.
You will be prompted with a familiar Compare form with the phrase changed to
suit the respective criterion you are comparing:
As you would notice, now we are comparing the relative value of alternatives with
respect to (wrt) Reliability in meeting the above Goal. Select the values with the
ones shown on the following Pairwise Matrix:
Please take note that these values do not represent the actual values used but
only for demonstration and make believe purposes only. Careful and balanced
judgment must be made for the results to be valid.
For Comfort
For Safety
Note: The writer assumes that all Proton Cars and KIA have the same technology for Safety but
inferior to Honda.
Final Note:
As you would notice, only 8 alternatives are allowed for this mode. This is not a
system constraint but a factor that attributes to using pairwise comparisons; as
for every C number of factors there would be C*(C-1)/2 comparisons needed to
be done.
4.7 Score
The Score is automatically calculated and a performance chart (on the right) is
updated for further analysis.
As can be seen, Honda City is the best overall despite having the highest sticker
price; with Waja a distant second, Gen2 third and Spektra last.
You can click Ranking->Ranking Table for more details.
As calculated weights by AHP are basically ratio scales, we can do some handy
calculations. You can evaluate the scoring according to one of the following
format:
4.8 Analysis
The software comes with many analytical functions, summarized as follow:
Name Description
Head-to-Head Compare two alternatives on a criteria to criteria basis in a horizontal
Analysis bar format
Compare two alternatives on a criteria to criteria basis in a spider chart
Spider Chart
format
Synthesis Priority preferences of criteria
Switch between Performance Analysis and Gap Performance Analysis;
Score/Gap Chart
criteria to criteria
New Score Chart Open New Score Chart.
Scores are recalculated according to actual global weights of
Weighted Score
contribution. Non-weighted scores are based on local performance
Chart
scores
Gradient
A separate form is displayed for what-ifs scenarios analysis
Analysis Tool
Weighted
Redisplay chart using weighted values
View
New Chart. To select different set of
New
alternatives pair
4.8.3 Synthesis
The priorities of the criteria are displayed when the button is pressed. It is sorted
in descending order. The top 80% of criteria contributing to the goal is marked in
red by the last criterion.
The current chart will instead show a chart showing differences from the ideal
values.
4.8.8 Series
This button shows the overall scores in a simple ascending format.
To define scale, select scale from main menu to get the Scale form as follows:
There are 3 columns to work from, namely: - criteria, scale and question
columns.
4.11.3 Menu
The menu commands are listed as follow:
Import Scale from other files. Function is not
Import Scale
available in this version
Create Generic Create scales automatically from current
Scales criteria but are not linked
Create Generic Create Generic Scales and automatically link to
Scale
Scales the criteria
New Create new scale
Register Register Scale
Save Scale locally. However, the scale is not
Save Scale
permanently saved in the file.
Link To Worksheet Link to Worksheet
Link to MapX. Function is not enabled in
Link Link to MapX
this version
Link to ADO Link to ADO Compatible Databases
Link To Data Link Criteria to Data kept in external database
Criteria
Unlink To Data Unlink criteria from database
Compare Relative Relative Pairwise Comparison
Data Data Comparison
1 Assign Selected Scale to Selected Criterion
Assign All Assign Selected Scale to All Criteria
Unassign Unassign selected criterion
Notes Function not available in this version.
Edit Levels Change Level names
Edit Ranges Change Range specifications
Actions
Edit Question Change Question associated with current scale
Edit Scale
Change Scale Description.
Description
Help Function not available online in this version
When this button appears, the text box is in
Save Button
Edit mode. Click this button to save.
Range Text <= Select <= mode as range operator
Select >= mode as range operator. Function
>=
not available in this version.
To terminate the ranges, the last value on the right (topmost green box) is to be
made a 0:
Check the Edit Ranges checkbox to terminate the edit ranges mode.
Note: The Min and Max are not used in this version.
4.12 Linking to Data in Worksheets and databases
The software allows rating data to be read from Worksheets and databases at
the same time. At the time of print, it can read also data from datasets of spatial
database, and results are output to a map. However, in this version, the function
is not made available.
Once the function is selected, you will be prompted with one of the following
forms depending which function is selected:
Select the appropriate files. You are allowed to specify only one spreadsheet and
database at a time.
Once these linkages have been defined, you can now link the criteria to the
specific fields of these data sources.
The above is shown for Reliability criterion. Select the field and click Link button
to complete the process. Continuously, doing it for other criteria that need to be
linked before you proceed to ranking table. Please take note you can only do it
once the scale levels have been defined.
Practically, not all data are needed in the worksheet. Only those data that are
numerical or real values can be stored in the Excel worksheet or database.
You can also use indexed value of the scale; although care must be taken that
the index exists, otherwise a zero rating is returned instead. For text scale, like
“Average” must be stored exactly as it is spelled.
For Microsoft Access (*.mdb files) files, the linked table must be named “bSure”
and “OPTIONS” field is mandatory and must be correctly spelt.
During ranking, all the data linked to worksheet or database will be automatically
filled in.
For unlinked criteria, the user is required to go through the combo box list to
select the rating.
Another important feature of the software is that once all the criteria have been
compared, enabling or disabling a criterion will automatically recalculate the
overall scores.
5 References