GRAPHFORECAST.RPF

GRAPHFORECAST.RPF is an example of graphical display of forecasts.

It uses ESMOOTH (exponential smoothing) to forecast Japanese total retail sales. The data are monthly from 1960:1 to 2006:12.

cal(m) 1960:1

open data oecdsample.rat

data(format=rats) 1960:1 2006:12 jpniptotrs

This uses ESMOOTH with an automatically selected model to forecast 24 periods out-of-sample. The forecasts are put into the series JPNFORE.

esmooth(trend=select,seasonal=select,forecast=jpnfore,steps=24) jpniptotrs

This inserts the final actual value of the data at the beginning for the JPNFORE series:

set jpnfore 2006:12 2006:12 = jpniptotrs

Finally, this forecasts the actual data from the final three years (2004:1 to 2006:12), with the forecast data starting at 2006:12 (which was the final actual) through 2008:12. A vertical grid line is added at the 2006:12 entry to show the end of the data.

graph(header="Forecasts of Japanese IP",grid=(t==2006:12)) 2

# jpniptotrs 2004:1 2006:12

# jpnfore 2006:12 2008:12

Full Program

cal(m) 1960:1

open data oecdsample.rat

data(format=rats) 1960:1 2006:12 jpniptotrs

* Use ESMOOTH with automatically selected model to forecast 24 periods

* out of sample. Add the final actual to the period before the forecasts

* start and put a grid line at the last actual data point.

esmooth(trend=select,seasonal=select,forecast=jpnfore,steps=24) jpniptotrs

set jpnfore 2006:12 2006:12 = jpniptotrs

graph(header="Forecasts of Japanese IP",grid=(t==2006:12)) 2

# jpniptotrs 2004:1 2006:12

# jpnfore 2006:12 2008:12

Graph

Output (from the ESMOOTH choosing the model)

Exponential Smoothing for Series JPNIPTOTRS

Model Selection

TREND SEASONAL SumSquares SBC

None None 3668.45317 1062.4

None Additive 3827.11790 1092.6

None Multiplicative 4182.33142 1142.7

Linear None 551.21387 -0.3

Linear Additive 541.19230 -4.3

Linear Multiplicative 577.28163 32.1

Exponential None 630.73239 75.7

Exponential Additive 642.43559 92.4

Exponential Multiplicative 650.68991 99.6

Model with TREND=Linear , SEASONAL=Additive

Alpha (level) 0.631238

Gamma (trend) 0.285467

Delta (seasonal) -0.045443