Self-organizing Inductive Modeling for Probabilistic Electricity Price Forecasting

Abstract

Self-organizing, inductive modeling (GMDH) is a proven and powerful data-driven modeling technology for solving ill-posed modeling problems as found in energy forecasting and other complex systems. It develops analytical, optimal complex, predictive models, systematically, from sets of high-dimensional noisy input data. The paper describes the implementation of rolling self-organizing modeling, exemplarily, for the Global Energy Forecasting Competition 2014 (GEFCom2014) probabilistic electricity price forecasting track using the KnowledgeMiner INSIGHTS inductive modeling tool out-of-the-box.