The FORECAST feature in prior releases was implemented for use in a RECAP command. However, the use of RECAP imposed limitations
on placement of the FORECAST field in the output and use of sort fields.
Implementing FORECAST in a COMPUTE command eliminates these limitations and enables you to place the FORECAST calculation
in a Master File. For the COMPUTE version of FORECAST, each type of calculation has its own version of the FORECAST function.
The calculations you can make to identify
trends and forecast values are:
- Simple moving average (FORECAST_MOVAVE). Calculates
a series of arithmetic means using a specified number of values
from a field.
- Exponential moving average. Calculates
a weighted average between the previously calculated value of the
average and the next data point. There are three methods for using
an exponential moving average:
- Single exponential smoothing (FORECAST_EXPAVE). Calculates
an average that allows you to choose weights to apply to newer and
older values.
- Double exponential smoothing (FORECAST_DOUBLEXP). Accounts
for the tendency of data to either increase or decrease over time
without repeating.
- Triple exponential smoothing (FORECAST_SEASONAL). Accounts
for the tendency of data to repeat itself in intervals over time.
- Linear regression analysis (FORECAST_LINEAR). Derives
the coefficients of a straight line that best fits the data points and
uses this linear equation to estimate values.
For more information, see the Creating Reports With WebFOCUS Language technical content.