Model Method - Profiling

The Profiling method implements the usage pattern and weather response for the forecast from a Profile Source. This source can be a Meter, a Meter Group, or a Model, but must cover the forecast period. The Model also has a Data Source, which is target data to which the forecast will be scaled. The Data Source can also be a Meter or a Meter Group.

The Profiling model specification options uses the following settings:

  • Data Source. This is identified in the Common Settings section of the General Tab. When selecting the source, first specify the MPI and UOM for the model. These inputs are used to filter the list of sources available to be installed as the Data Source. Sources will be limited to meters or meter groups with a frequency that is as small as the model frequency or smaller. For example, an hourly model can have a 15-minute data source.
  • Profile Source. The sources available include Meters, Meter Groups, and Models. For Meters and Meter Groups to be used as a profile source, they must contain forecast values as well as historical data. Clicking the search icon to the right of the Profile Source list will open a search dialog to select the type of the source and the source itself.
  • Adjust to History. This step applies when there are data available for both the Profile Source and the Target variable in a recent historical period (the overlap period).
  • History Days. If Adjust to History is checked, the user defines the number of days to be used for the overlap calculation. For example, if the number of History Days is set to 2, then the most recent two (2) days of historical data by day-of-the-week will be used. For example, with at setting of 2, the most recent two Sundays will be used to adjust the Sunday shape forecasts, the most recent two Mondays will be used to adjust the Monday forecasts, and so on.
  • Use Holidays. This option appears in the General Tab of the Model Properties dialog. If this box is checked, holidays in the historical data are ignored when computing unitized shapes for the Data Source and Profile Source variables. In generating the forecasts, if a forecast day is a holiday, the shape adjustment for Sundays is applied to that day.