Vous êtes sur la page 1sur 2

Exponential Smoothing

GRETL does not provide a way for you to forecast using exponential smoothing as described in Chapter 4
(starting on page 119). Your instructor has developed Excel templates to assist you in performing the
exponential smoothing forecasts for your companies.
There are four principal exponential smoothing techniques:
1.
Single Exponential Smoothing (uses one parameter: ).
2.
Double Exponential Smoothing-Holts Method (uses two parameters: and ).

3, 4.

Triple Exponential Smoothing-Winters Method. Winters method has two


approaches in how it treats seasonality. One handles additive seasonality; the
other multiplicative seasonality (both use three parameters: , , and ). Your
textbook only discussed multiplicative seasonality.

There are four Excel template files on the course page for you to download and use:
Smoothing_single.xlsx
Smoothing_double.xlsx
Smoothing_triple_add.xlsx
Smoothing_triple_mult.xlsx
Since you will be using each template twiceonce for each of your companiesyou probably
will want to save the files using different names. I suggest that you adopt the following naming
protocol: [Company Name]_[Smoothing Method].xlsx where smoothing method is abbreviated
ses
for single exponential smoothing
des
for double exponential smoothing
tesa for triple exponential smoothing additive
tesm for triple exponential smoothing multiplicative
Example: if your company is Apple, name your single exponential smoothing file with the Apple
sales data as follows: Apple_ses.xlsx
The files are all similar in their layout structure, and you have to do the SAME THINGS in each file to
make them work.
1. Open YOUR Excel file that contains the Real Sales Data for your company.
2. Open one of Exponential Smoothing template files.
3. In the template file, you will see three shaded areas (yellow, pink, and blue)
a. Fill in your company name in the yellow shaded cell
b. Fill in the number of observations of your real sales data in the pink cell (40 is the default).
c. Copy your Real Sales data (from your spreadsheet) and PASTE VALUES into the blue shaded
column labeled Actual Value. Make sure to paste the data so the first observation is shown
next to the number 1 and that that last observation in the C column exactly corresponds to
the number of observations you entered in the pink shaded cell (from part (b) above).
NOTES: (1) The number of observations should continue to count four (4) periods beyond the
observations in your data. This is the forecast area. Do NOT make any entries in that area.
If your data does not correspond to the number of periods you indicated when filling in the

yellow cell from part (b), correct the number in the pink cell. (2) The blue shaded part of
column C where you paste in your real sales data extends for 100 data points. This is the
maximum this template will handle. Since most of you have fewer observations than that,
your data will end and the blue strip MUST contain blank cells below your last data point.
(3) DO NOT Change the content of any cells that are NOT highlighted unless you are using
the sliders or designating a seed value (the fourth option in the Seed Value pull-down menu).
(4) The template will have error statements until the data has been entered and the number of
observations is designated correctly. If you do not enter this information correctly, your
forecasts will be wrong!!!
Slider(s): Depending on the forecasting method, you will see 1 3 control sliders in the upper left
portion of the template. These allow you to control the parametric values of the forecasting
process.
Charts:

You should add three charts to each template. These could not be prepared as part of the
template since the charts depend on the number of observationswhich vary for each of you.
ACTUAL and FORECASTED VALUES: This is a simple chart. Select the two cells with the
labels Actual Values and Forecast Values and all of the data cells below them (there
should be exactly 4 blank cells at the bottom of the Actual Value column that you have
selected.) Click on the Insert Tab and choose Line Chart. Click on a chart type, and the chart
will be inserted into your spreadsheet.
ERROR VALUES: Create this chart in the same manner only use the single column of Error
data. When selecting the data, make sure to include the Error label cell as the top cell in
your selected area.
CORRELOGRAM: Select the data in the three columns (and their labels) that are labeled
Autocorrel Values (AC), Upper 95% Value, and Lower 95% Value. Click on the Insert
Tab and select Column Chart. Click on the side-by-side (not stacked!) 2-D chart choice, and
your chart will be inserted. This one will need to be cleaned upYou will be shown how in
class.

Vous aimerez peut-être aussi