Preparing Time Series Data

This topic describes the general requirements for getting source data into SAS Visual Forecasting. For further information about time series data, see Understanding Time Series Data.

Requirements for Time Series Data

SAS Visual Forecasting projects require input data that is in time seriesan aggregation of transactional data into specified time intervals and sorted according to unique combinations of the default attributes (BY variables) format. Each observation or row in the table should be using a SAS date or datetime format.

Some time series data is equally spaced. That is, successive observations are a fixed time interval apart, and the data can be described by a single interval, such as daily, every three weeks, or twice a month. You might already have this time series data, or you might have time-stamped data at irregular intervals. You can use the accumulationeither of two processes that are used to convert a time series. (1) Accumulation converts a time series that has no fixed interval into a time series that does have a fixed interval (such as hourly or monthly). (2) Accumulation converts a time series that has a fixed interval into a time series with a lower frequency time interval (such as hourly into daily). Accumulation combines data within the same time interval into a summary value for that time period. options in SAS Visual Forecasting to convert the time-stamped data into a time series.

Your time series data must meet the following requirements:

The data set contains the dependent variable that needs to be modeled and forecasta numerical prediction of a future value for a specified time period for each unique combination of BY variable values. If multiple variables need to be forecast, you can set that up in separate projects. Multiple projects can be run using the same source data set.
The data set contains a variable that is in a SAS date or datetime format. When SAS Visual Forecasting creates a project, the data is sorted by the time variable so that the observations are in order according to time. For more information, see Date, Time, and Datetime Constants in SAS Programmer’s Guide: Essentials.
The data is loaded into a table in Cloud Analytic Services (CAS). You can use SAS Data Explorer to import data into CAS, or to load a SAS data set into CAS. See SAS Data Explorer: User’s Guide for more information.
The data must have a time series that includes at least ten observations.
The variable names in the data set cannot use any of the restricted names. See Reserved Names for Variables and Models for a complete list.

Note: The names of the variables cannot match any of the reserved variable names that are used in the output data set. For more information, see Reserved Names for Variables and Models.

For general information about working with time series data, see Working with Time Series Data in the SAS/ETS User’s Guide.
For more information about creating time series data from transactional datatimestamped data collected over time at no particular frequency. Some examples of transactional data are point-of-sale data, inventory data, call center data, and trading data., see "The TSMODEL Procedure" in the SAS Visual Forecasting: Programming Guide.

Converting Dates from Character to SAS Date Format

Often a data source might have the date or date and time in character format. For example, consider a variable with the name Sale_Date with this construction:

ddd, yyyy mmm dt hh:mm

ddd = 3-letter day of the week
yyyy = 4-digit year
mmm = 3-letter month
dt = 1- or 2-digit day of the month
hh = 1- or 2-digit hour in 24-hour notation (0–23)
mm = 2-digit minute

An example of an observation in this format is “Mon, 2017 May 8 8:08”.

To convert this character string into a SAS date, you would need to use some SAS functions in a SAS DATA step. The following functions extract the year, month, and date portions of the Sale_Date variable. The TRIM function concatenates them into a single string that can be read into a new variable, Project_Date, using the SAS date9. informat.


Year = substrn(Sale_Date,6,4);                   /* 1 */ 
Month = substrn(Sale_Date,11,3);                 /* 2 */ 
Date = substrn(Sale_Date,15,2);                  /* 3 */ 
Datechar = trim(Date)||trim(Month)||trim(Year);  /* 4 */ 
Project_Date = input( Datechar , date9.);        /* 5 */ 
Drop Year Month Date Datechar;                   /* 6 */ 
Format Project_Date date9.;                      /* 7 */

Extract the year starting at the sixth position.
Extract the month starting at the 11th position.
Extract the date starting at the 15th position.
Trim trailing spaces from each new variable, concatenate the three variables into ddmmmyyyy format.
Read in date string as SAS date9. informat.
Drop temporary variables from the final data set.
Set date format for time variable.

Using the code in this example, the string Mon, 2017 May 8 8:08 is converted to the SAS numeric date of 20947. Use the FORMAT statement to format the variable so that SAS Visual Forecasting will properly detect the data when you create a project.

SAS administrators can also use SAS Data Studio to convert string data to a SAS date or datetime format.

Reviewing Your Variable Names

There are some strings that are used by SAS Visual Forecasting that can cause errors when they match the name of variables in the project data set. See Reserved Variable Names for a complete list so that you can ensure that you do not have any variable names that might cause conflicts.

Last updated: March 16, 2026