Master's Culminating Experience
Dayton Power and Light commissioned this capstone project to test DP&L’s hypothesis that individual forecasting models for each of its different customer classes when combined will generate a better forecast of total load than their current practice of using one model that does not make a distinction between the electrical use of different customer classes. This hypothesis was based on the observable fact that a change in weather does not create a change in electrical consumption equal across all customer classes. Forecasting models were created for each customer class within the data limitations. A forecasting model for total load was also created. The forecast of total load generated by the sum of the customer class models was consistently more biased and less accurate than the one total load model. Therefore, the conclusion of this project is that DP&L should continue to use only one model that does not identify who uses the electricity.
Gleason, A. J.
(2009). An Investigation into the Forecasting Methods of Dayton Power and Light. .