Model Settings

On the Model Specification Tabs, there are three or four options allowed, depending on the modeling concept. These are:

  • Use Average Values. This option will use the average values by month over the estimation period to generate the forecast.
  • Use Fixed Ratio. This option is only available on the tabs labeled Customer CF, GSP CF, TTSL CF, CA CF, and Min Ratio. It allows the user to enter a fixed ratio multiplier that will be used to generate the forecast.
  • Use Standard Model. This option allows the user to select a subset of variables to include in the model, and also set some parameters.
  • Use Custom Model. This option allows the user to specify a MetrixND template file to be used for estimation.
  • SAE Option. This is an additional option for when Use Standard Model is checked. This is only available for Energy, NCP, and HLH models.

The specifics of these options are described below.

Use Average Values. When Use Average Values is selected, there are no calculation options or XDrivers. Average values by month are calculated over the estimation period and are used as the forecast values. Error statistics are calculated from the deviations of actual values around average values by month. The coefficients and supporting statistics are presented with one row for each month.

When Seasonal is checked, the model uses a seasonal average instead of a monthly average as described above.

Use Fixed Ratio. This option is available for the coincidence factor (CF) and minimum ratio (Min Ratio) models. When Use Fixed Ratio is selected, the average factor over the estimation period will be computed and this value will be displayed in the test box to the right of the option. The user may then edit this value. A checkbox will allow the user to specify that this value should not be changed when the model is re-estimated, effectively “locking” the user entered factor for the model. Error statistics are calculated from the deviations of monthly actual values around the fixed ratio.

Use Standard Model. When Use Standard Model is selected, a list of modeling options is presented to the user. These modeling options vary across model types.

Use Custom Model. When Use Custom Model is selected, the user must select a MetrixND template to use for the model. These templates are managed using the Manage Templates Dialog. This option includes a text box that will provide the list of all templates configured for a specific Long-Term modeling concept (Energy, NCP, …) . Once a template is selected, the user can export a MetrixND project file (.ndm) that is populated with data for the model by pressing the Export… button to the right of on the template selection box. This brings up a Save As dialog allowing the user to specify the file name and destination drive for the export.

Including X Drivers and Model Drivers. With the Standard Models and Custom Models, the user can specify a list of X Driver variables and Model variables to include in the model. The list of variables included in the model appears toward the bottom of the specification form.

To include an X Driver, press the Add XDriver button. This will open a Select XDriver dialog. This dialog supports sorting and filtering of the list of X Drivers. Find the desired variable and press the Use Selected button. This will add the selected XDriver to the list of explanatory variables in the model.

To include a model predicted from another model as an explanatory variable, press the Add Model button. This will open a Select Model dialog. This dialog supports sorting and filtering of the list of models. Find the desired model and press the Use Selected button. This will add the predicted value from the selected model to the list of explanatory variables in the model.

SAE Option. With Standard Models the SAE Option allows the user to include Statisticall Adjusted End Use inputs in the model. The SAE variables that are used are XHeat, XCool and XOther.

  • Monthly XHeat variables are interacted with HDD variables if Include HDD is checked, and also with TrendHDD variables if Include TrendHDD is checked.
  • XCool is interacted with CDD variables if Include CDD is checked, and also with TrendCDD varialbes if Include TrendCDD is checked.
  • XOther appears by and does not interact with other variables.

There is an additional option to exclude the constant term. This forces the SAE variables explain the monthly sales levels through time without an underlying base or constant value. If Include Monthly Binaries is checked and Exclude Constant is checked one of the monthly binary variables will be excluded from the model.

An Elasticity Adjustment option is available and is defaulted to 1.0. This numerical value modifies the Scale Variable elasticities in the SAE energy calculations. A value of .7 will reduce the scale elasticities by from the assigned SAE Set by 30%, which is appropriate if the Customer energy growth is below the growth of the economic scale drivers. A diagnostic value is provided by the system by pressing the Calculate button. This will give a value greater than 1.0 if the Customer energy growth is greater than the SAE model growth and a value less than 1.0 if Customer growth is slower than the SAE model growth.

Adjustments and Overrides. The adjustment and override settings define Adders, Multipliers, and Override values that are applied after model calculations are complete. Multi-value overrides are implemented using monthly XDriver variables.

  • XDriver Adder. Monthly data values for Adder variables will be added to the Forecast.
  • XDriver Multiplier. Monthly data values for Multiplier variables will be used multiplicatively to scale the forecast.
  • XDriver Override. Monthly data values for Override variables will be used in place of the model generated forecasts.

Single-valued factors can also be entered in the Factor Options box. To activate this approach, click the Use Factor checkbox, enter the data value for the factor, and pick the factor type. The three single value factor types are:

  • Adder. This value will be added to the forecast each month..
  • Multiplier. This value will be used multiplicatively to scale the forecast each month.
  • Override. This value will be used to replace the model generated forecast each month.