The Series

---

At the very beginning of the research, monthly demands of 26 items were available for demand analysis. The number of monthly demands available at that time and the summary statistics of the 26 items are shown in Table 1. From Table 1, it can be seen that the items cover a wide range of mean demand and demand variation. The mean demand ranges from 7 units to 6,627 units while the coefficient of variation ranges from 18% to 121%. In order to select a sample of items which can represent the total set of data series on forms usage at the bank for further analysis, the followings were considered:

1. The range of the mean demands included by the sample should be wide enough to make the sample representative; however, the items in the sample should also be of economic significance.

2. The sample should be able to reveal the variations of monthly demands of forms.

3. Data availability.

Table 1: Summary statistics of the 26 items

Item	Number of Monthly Demands	Mean	Standard Deviation	Coefficient of Variation (%)
ATM106	35	27.82	20.04	72.04
COVRPBN	48	30.20	7.79	25.80
CPD092	35	1727.86	387.69	22.44
CUA032	35	688.03	195	28.34
CUA085[a]	48	1376.18	488.79	35.52
CUA778[a]	48	3930.67	928.94	23.63
E057	35	245.33	65.59	26.74
PB0011	35	81.11	14.70	18.12
PB0041	35	7.00	3.07	43.86
REM012	35	735.34	170.09	23.13
REM037[a]	48	1336.10	526.85	39.43
REM061[a]	48	843.33	235.26	27.90
REM780	35	1263.80	444.13	35.14
SAV012[a]	48	2679.51	837.39	31.25
SAV013[a]	48	2720.63	698.27	25.67
SAV091[a]	48	118.44	89.87	75.88
SAV092	35	136.23	143.79	105.55
SAV133	35	68.56	82.97	121.02
SAV139	35	82.97	27.79	33.49
SAV140	35	94.17	27.95	29.68
SAV739[a]	48	332.56	78.01	23.46
SAV740[a]	48	170.80	47.09	27.57
SAV763[a]	48	6627.10	1476.98	22.29
TMD095	48	2670.40	1597.29	59.82
TMD096	48	1830.88	1300.60	71.04
TMD098	13	1527.00	1077.99	70.60
[a]Items selected for further analysis

In the consideration of economic significance, the management suggested that an item having a mean monthly demand less than 100 is of little significance. Based on this reason, 7 items were excluded from further analysis. In the remaining 19 items, the coefficient of variation of item SAV092 was 105.55% and was significantly higher than those of the others (22.29% to 75.88%). Hence, item SAV092 was also excluded. This left 18 items for further consideration of their data availability. In order to have sufficient data for forecasting model building, combination weight estimation and testing, it was decided that 60 monthly demand data for each item were required. This implied that items having 35 monthly demand data would require another 25 months' data while items having 48 monthly demand data would require another 12 months' data. Obviously, waiting for another 25 months for collecting the data was not feasible. Hence, it was decided that only those items with 48 monthly demand data were selected for the research. There were 12 items. After a few months, it was found that 2 items in the 12 became obsolete and would never be used by the bank; these two items were items TMD095 and TMD096. As a result, 10 items were later used by this research for further analysis. These l0 items are marked by an asterisk in Table 1. From Table 1, it can be seen that the 10 items selected cover a wide range of mean monthly demands from 118.44 to 6,627.10. This wide range is expected to be representative to the different demand levels of the main consumption of the bank forms. The range of the coefficients of variation of the 10 items is from 22.29% to 75.88%. It can be seen from Table 1 that 23 items out of the original 26 items are having their coefficients of variation within this range. Hence, the 10 items selected should also be a representative sample of the variations of monthly demands of the bank forms. After the selection of the 10 items, 12 more monthly data for each item were then collected. Monthly demand data for these 10 items was available afterwards for a reasonably long period of 60 months, from January '91 to December '95.

---