Seasonality, also known as cyclical data, means that data for some unit of time repeats in a regular pattern. For example, if you have 24 monthly data points, and the data has peaks every December, the seasonality (repeating pattern) has a period of one year or 12 months.
Use the Data Attributes panel of the Predictor wizard to perform the following tasks:
Specify time-period and seasonality information for historical data
Define events that influenced data values
Apply optional screening to replace missing values and locate and replace data outliers
Specifying Time Periods and Seasonality
To specify time periods and seasonality:
Display the Data Attributes panel of the Predictor wizard.
To display Data Attributes, click Next in Input Data or click Data Attributes in the navigation pane of the Predictor wizard.
For Data is in, identify the time period for the data.
For example, if the data points represent monthly numbers, select months.
For Seasonality, indicate whether the data is seasonal:
AutoDetect—Uses statistical algorithms to determine whether the data is seasonal. Findings are displayed in a statement to the right of the list box.
Non-seasonal—Indicates that data is treated as non-seasonal; seasonal methods will not be applied.
Seasonal—Indicates that seasonal and non-seasonal methods are used by default. You must have at least two seasons (complete cycles) of data to use the seasonal methods.
Optional: If you are analyzing more than one data series click View Seasonality to review seasonality for each series.
For more information, see Viewing Historical Data by Seasonality.
Specify how to treat missing values and outliers (historical values that differ extremely from other values):
Select Fill-in missing values to fill in missing data values using settings in the Data Screening Options dialog.
Select Adjust outliers to eliminate extreme values from the data before the time-series forecasting methods are run.
Notice that the default values (filling in missing values but not adjusting outliers) are appropriate for most cases. For details, see Viewing Screened Data.
Optional: Click View Events to define and manage events—time periods where data could have been affected by unusual occurrences such as promotions, weather, holidays, and strikes.
If you have defined an event, you can select Include Events to incorporate event definitions into forecasts. For details, see Viewing and Managing Events.
Optional: Click View Screened Data to view a chart of filled-in values and adjusted outliers. For more information, see Viewing Screened Data.
When settings are complete, click Next to open the Methods panel.