FORECASTING INVENTORY

FORECASTING INVENTORY

DistCo, a large warehouse service company in the San Francisco Bay Area, stores pharmaceutical products for customers while they are in transit to local retailers. DistCo can store a maximum of 280,000 cases of products at its present facility. Because their business has been growing, the company’s management wonders if they should acquire other warehouses in 2001. The materials specialist has accumulated the following inventory data:

Inventory (thousand
Year Quarter Period of cases)
______________________________________
1996 1 1 176
2    2    134
3    3    186
4    4    195
1997    1    5    189
2    6    157
3    7    195
4    8    211
1998    1    9    205
2    10    180
3    11    212
4    12    229
1999    1    13    223
2    14    192
3    15    234
4    16    248
2000    1    17    239
2    18    217
3    19    271
4    20    284
_______________________________________

1. Select the best forecasting model and justify why it is better than other model.
2. What quarterly inventory is to be expected in each quarter of 2001?
3. Should DistCo acquire more warehouse capacity in 2001?
4. What are the potential factors that may affect the forecast accuracy of the model you have selected?

Maybe should use the DS for Windows

DS for Windows is a software to help run time-series forecasting. The following are steps to install and use DS for Windows (not for Mac OS)

Down load these four files in one folder (make sure no other file in this folder)
Install DS for Windows by clicking the SETUP file
Open DS for Windows and select “forecasting” module
Go to “File” menu à New à Time series analysis
Set number (20 in this case) of past period and key-in the data
No naïve, so go to moving average. Select n: number of period to be averaged. Then solve.
Explain the forecast result and error analysis. How to get forecast for next four periods? Go through weighted moving average model
Go through exponential smoothing with or without trend adjustment How to select the best alpha and Beta? By what criteria?
Go through linear projection or linear regression. They will get the same a and b.
Go through the multiplicative decomposition carefully and explain how to get the basic as well as the seasonal factors. Choose # (4 in this case) of season and average method.
The unadjusted part is actually from linear trend projection.