Utilizing time series analysis to forecast long-term electrical consumption
Danny Modlin · NC Digital Online Collection of Knowledge and Scholarship (The University of North Carolina at Greensboro) · 2009
Developing an accurate forecast model for the amount of power consumed will include such factors as time of day, day of the year, the weather, among many others. Based upon these given factors, current models use a neural network approach to forecast in the very near future. For the purpose of business operations, this model should be accurate for predicting power usage at least six months into the future. Using regression with time series analysis, the goal is to build a model that re°ects systematic movements in the data and predict them so errors would be more or less random and minimized.