A novel data-characteristic-driven modeling methodology for nuclear energy consumption forecasting
•A novel data-characteristic-driven modeling methodology is proposed.•The methodology formulates forecast model based on data’s own data characteristics.•Two steps are involved: data analysis and forecast modeling.•Relationships between data characteristics and forecasting models are discussed.•Empi...
Saved in:
Published in | Applied energy Vol. 128; pp. 1 - 14 |
---|---|
Main Authors | , , |
Format | Journal Article |
Language | English |
Published |
Kidlington
Elsevier Ltd
01.09.2014
Elsevier |
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | •A novel data-characteristic-driven modeling methodology is proposed.•The methodology formulates forecast model based on data’s own data characteristics.•Two steps are involved: data analysis and forecast modeling.•Relationships between data characteristics and forecasting models are discussed.•Empirical results statistically verify the effectiveness of our novel methodology.
Due to the unique features of nuclear energy market, this paper tries to propose a novel data-characteristic-driven modeling methodology based on the principle of “data-characteristic-driven modeling”, aiming at formulating appropriate forecasting model closely in terms of sample data’s own data characteristics. In the novel data-characteristic-driven modeling methodology, two steps are mainly involved, i.e., data analysis and forecasting modeling. First, the sample data of nuclear energy consumption are thoroughly investigated in order to capture the main inner rules and hidden patterns driving the data dynamics, in terms of data characteristics. Second, the corresponding forecasting model is accordingly formulated and designed based on these data characteristics. For illustration and verification purposes, the proposed methodology is implemented to predict the nuclear energy consumption of USA and China. The empirical results demonstrate that the novel methodology with the principle of “data-characteristic-driven modeling” strikingly improves prediction performance, since the models elaborately built based on data characteristics statistically outperform all other benchmark models without consideration of data characteristics. This further confirms that the proposed methodology is a very promising tool in both analyzing and forecasting nuclear energy consumption. |
---|---|
Bibliography: | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23 |
ISSN: | 0306-2619 1872-9118 |
DOI: | 10.1016/j.apenergy.2014.04.021 |