The FORECAST Procedure

Functional Summary

Table 17.1 summarizes the statements and options that control the FORECAST procedure.

Table 17.1: FORECAST Functional Summary

Description

Statement

Option

Statements

   

specify model and data set options

PROC FORECAST

 

specify BY-group processing

BY

 

identify observations

ID

 

specify the variables to forecast

VAR

 

Input Data Set Options

   

specify the input SAS data set

PROC FORECAST

DATA=

specify frequency of the input time series

PROC FORECAST

INTERVAL=

specify increment between observations

PROC FORECAST

INTPER=

specify seasonality

PROC FORECAST

SEASONS=

specify number of periods in a season

PROC FORECAST

SINTPER=

treat zeros at beginning of series as missing

PROC FORECAST

ZEROMISS

Output Data Set Options

   

specify the number of periods ahead to forecast

PROC FORECAST

LEAD=

name output data set to contain the forecasts

PROC FORECAST

OUT=

write actual values to the OUT= data set

PROC FORECAST

OUTACTUAL

write confidence limits to the OUT= data set

PROC FORECAST

OUTLIMIT

write residuals to the OUT= data set

PROC FORECAST

OUTRESID

write standard errors of the forecasts to the OUT= data set

PROC FORECAST

OUTSTD

write one-step-ahead predicted values to the OUT= data set

PROC FORECAST

OUT1STEP

write predicted, actual, and confidence limit values to the OUT= data set

PROC FORECAST

OUTFULL

write all available results to the OUT= data set

PROC FORECAST

OUTALL

specify significance level for confidence limits

PROC FORECAST

ALPHA=

control the alignment of SAS date values

PROC FORECAST

ALIGN=

Parameters and Statistics Output Data Set Options

   

write parameter estimates and goodness-of-fit statistics to an output data set

PROC FORECAST

OUTEST=

write additional statistics to OUTEST= data set

PROC FORECAST

OUTESTALL

write Theil statistics to OUTEST= data set

PROC FORECAST

OUTESTTHEIL

write forecast accuracy statistics to OUTEST= data set

PROC FORECAST

OUTFITSTATS

Forecasting Method Options

   

specify the forecasting method

PROC FORECAST

METHOD=

specify degree of the time trend model

PROC FORECAST

TREND=

specify smoothing weights

PROC FORECAST

WEIGHT=

specify order of the autoregressive model

PROC FORECAST

AR=

specify significance level for adding AR lags

PROC FORECAST

SLENTRY=

specify significance level for keeping AR lags

PROC FORECAST

SLSTAY=

start forecasting before the end of data

PROC FORECAST

START=

specify criterion for judging singularity

PROC FORECAST

SINGULAR=

limit number of error or warning messages

PROC FORECAST

MAXERRORS=

Initializing Smoothed Values

   

specify number of beginning values to use in calculating starting values

PROC FORECAST

NSTART=

specify number of beginning values to use in calculating initial seasonal parameters

PROC FORECAST

NSSTART=

specify starting values for constant term

PROC FORECAST

ASTART=

specify starting values for linear trend

PROC FORECAST

BSTART=

specify starting values for the quadratic trend

PROC FORECAST

CSTART=