Load Forecasting and Statistical Services

GDS provides the statistical expertise required to develop forecasting models that are theoretically and statistically sound.

Load Forecasting and Statistical Services

John Hutts

Utilities require reliable load and energy forecasts to support various planning functions into today’s environment. Energy consumption is influenced by a multitude of factors, and it is imperative that these factors are captured in the models used in developing load forecasts. Additionally, forecasts are often scrutinized by regulators and intervenors during regulatory proceedings, by power supply bidders during proposals, or from potential lenders in loan applications. Whatever the purpose, GDS maintains the knowledge and resources necessary to provide the forecasting services needed by our clients.


GDS provides the statistical expertise required to develop forecasting models that are theoretically and statistically sound. Our models not only measure the impacts of many determining factors of energy consumption, but they also have withstood examination by outside entities that review and evaluate the models. GDS tailors a scope of work that is best suited for a client’s specific needs and budget. We develop day-ahead forecasts for resource scheduling, monthly forecasts for budgeting and near-term financial planning, and annual forecasts for long-term power supply and resource planning. We use the following state of the art forecasting techniques:

Multiple Regression Models – Simple multiple regression models provide a low-cost means for quickly quantifying the impacts of a limited number of factors on energy consumption and peak demand.

Statistically Adjusted End-Use (SAE) Models – SAE models combine the benefits of both end-use and econometric models. The models are regression based and capture the impacts of housing/building characteristics, householder characteristics, appliance market shares and efficiencies, the economic outlook, retail prices, weather conditions, national energy standards, and energy efficiency programs.

Neural Networks – Neural networks address non-linear relationships with respect to forecasting hourly energy/water consumption, and as a result, tend to deliver a higher degree of accuracy than traditional forecasting methods.


Utility planning often requires information that is only available through customer surveys, onsite audits, or load research. The costs for collecting such information can be significant; therefore, it is important to draw statistically-reliable customer samples that are representative of the entire population. GDS has successfully managed all phases of such projects, including:

  • Sample design
  • Sample selection
  • Interpretation of results
  • Determination of statistical precision
  • Reporting


GDS provides support in regulatory proceedings with respect to review and evaluation of statistically-based analyses. From integrated resource plans to wholesale and retail rate applications, GDS provides the expertise needed to defend the statistical analysis behind a client’s filing or to review and evaluate analyses on behalf of intervening parties. Areas of expertise include:

  • Load forecasts
  • Load research studies
  • Weather normalization


John Hutts