ISPRS, Vol.34, Part 2W2, "Dynamic and Multi-Dimensional GIS", Bangkok, May 23-25, 2001 
387 
each of which assumes a different set of weight assigned to 
evaluation criteria. 
The weights used in this demonstration are given in Table 2. 
Various sensitivity analyses were undertaken to determine 
these. Weights have been assigned to the criteria at both levels. 
Priorities are then calculated using Direct Rating methods. All 
priorities range between 0 and 1. 
Table 2. Criteria weights 
Level 2 Criteria /Level 3 Attributes 
Weights 
Priorities 
Catchment area specific variables 
100 
0.667 
Banking population (age 15 & over) 
100 
0.222 
Population (over 15) growth rate 
75 
0.167 
Average annual family income 
75 
0.167 
Average age 
75 
0.167 
Total (small) business No. 
75 
0.167 
Catchment area 
50 
0.111 
Location specific variable 
50 
0.333 
Small business in 200m buffer zone 
50 
0.133 
Competitor's branch No. in 200m buffer 
75 
0.200 
Working population in 200m buffer 
100 
0.267 
Same bank's branches in 500m buffer 
50 
0.133 
Shopping centers Within 500m buffer 
100 
0.267 
3.6 Evaluating the branch alternatives with respect to the 
criteria 
Evaluation using SMART involves the following phases: 
• Construct an objectives hierarchy to specify the criteria to 
be considered 
• Determine the evaluation criteria at each level in the 
hierarchy 
• Build the value relations between the alternatives being 
considered 
• Assign criteria weights (the critical aspect in MCDM) 
• Implement rank procedures to generate decision scores for 
alternative and sensitivity analysis 
The multiple criteria evaluation process is realised using the 
CDP module, the functions of which include the enumeration of 
criteria preferences, selection of aggregation functions, the 
generation of decision scores, sensitivity analysis, and attribute 
contribution analysis. 
4. RESULTS 
The SDSS outlined above has been applied in a number of 
pilot studies to test its efficiency. The results reported here are 
based on a random sample of 49 Commonwealth Bank 
branches in the Sydney Metropolitan area in 1995, some of 
which had been closed by 2000. The test of the success of the 
SDSS is whether or not it can identify these as candidates for 
closure. 
The criteria listed in Table 2 are used in combination as a basis 
for deciding whether or not a particular branch is a candidate for 
closure. The criteria are sorted hierarchically by level. Two levels 
are used in this demonstration (Fig 2). 
Using these criteria and weights, the SMART model generates 
the decision score for each candidate bank (Table 3), the 
ordering of which is taken to be the preference ranking. 
Branches with a high score (eg. Caringbah, Kingsway, 0.681) 
are considered to be viable and are therefore not likely 
candidates for closure. On the other hand, branches with a low 
score (eg. Brookvale, Old Pittwater Rd., 0.178) are associated 
with a particular combination of weighted criteria suggesting that 
the branch might be considered as a candidate for closure. 
Table 3. Decision scores by SMART 
Branch Names 
Decision Scores 
Value 
Order 
ASQUITH, PACIFIC HWY* 
0.783 
1 
Caringbah, Kingsway 
0.681 
2 
Canley Heights, Canley Vale Rd 
0.489 
3 
Cabramatta, John St 
0.488 
4 
Bondi Junction, Newland St 
0.485 
5 
Campsie, Beamish St 
0.469 
6 
BELROSE, SHOPPING CNTR* 
0.448 
7 
Ashfield, Hercules St 
0.421 
8 
Bass Hill, Bass Hill Plz 
0.412 
9 
Bankstown, Old Town Cntr Plz 
0.411 
10 
Avalon Beach, Avalon Pde 
0.409 
11 
Auburn, Auburn Rd 
0.402 
12 
Balgowlah, Sydney Rd 
0.393 
13 
Balmain, Darling St 
0.393 
14 
Blacktown, Westpoint Marketwn 
0.383 
15 
Baulkham Hills, Stockland Mall 
0.367 
16 
Artarmon, Hampden Rd 
0.345 
17 
Blacktown, Campbell St 
0.339 
18 
BLACKTOWN, NEWLAND ST* 
0.335 
19 
Belmore, Burwood Rd 
0.328 
20 
BONDI, BONDI RD* 
0.324 
21 
Baulkham Hills, Old Northern Rd 
0.318 
22 
Arncliffe, Belmore St 
0.315 
23 
Bexley North, Bexley Rd 
0.302 
24 
Brighton-Le-Sands, Bay St 
0.300 
25 
Annandale, Booth St 
0.299 
26 
BERALA, WOODBURN RD* 
0.293 
27 
Botany, Botany Rd 
0.290 
28 
Bankstown, Bankstown Service Centre 
0.285 
29 
Auburn, Parramatta Rd 
0.275 
30 
BELLEVUE HILL, BELLEVUE RD* 
0.270 
31 
CANLEY VALE, CANLEY VALE RD* 
0.269 
32 
Beverly Hills, King Georges Rd 
0.267 
33 
Beecroft, Hannah St 
0.260 
34 
Bondi Beach, Hall St 
0.260 
35 
CAMMERAY, MILLER ST* 
0.257 
36 
Brookvale, Pittwater Rd 
0.256 
37 
Beaconsfield, Botany Rd 
0.250 
38 
CANTERBURY, CANTERBURY RD* 
0.249 
39 
BELFIELD, BURWOOD RD* 
0.248 
40 
Burwood, Belmore St 
0.228 
41 
ALEXANDRIA, MITCHELL RD* 
0.209 
42 
Ashfield, Ashfield Mall 
0.200 
43 
Brookvale, Old Pittwater Rd 
0.178 
44 
CAMPERDOWN, GROSE ST* 
0.176 
45 
BURWOOD, BURWOOD RD* 
0.160 
46 
CAMPERDOWN, PARRAMATTA RD* 
0.152 
47 
CAMPERDOWN, BOOTH ST* 
0.127 
48 
BURWOOD, ELSIE ST* 
0.082 
49 
* indicates branches closed by 2000 
The results are very encouraging. A comparison of the branches 
identified by the model as likely candidates for closure with 
those that actually have been closed since 1995 indicates 
clearly that most branches with low scores have been closed. 
For example, of the 14 branches with score less than 0.260 in 
the SMART model, 9 have closed. There are however 
exceptions: some banks have been closed which the model 
suggested are viable