Short-term natural gas consumption forecasting from long-term data collection

The development of natural gas consumption forecasting tools is an important application of forecasting models. Plenty of research efforts have already been made in this area. However, the datasets used in these works could often not be published and used by other researchers. This complicates further research and the comparison of forecasting methods. In this work, we address this issue by the creation of a new dataset. We have taken into account state-of-the-art research works and included many data features that were previously proven to have a significant impact on the precision of the model. A forecasting methodology suitable for the evaluation of statistical and machine learning algorithms used in the time series forecasting domain is proposed to validate the high usability of the new dataset. The results of the application of the methodology and their discussion are included. Moreover, we made this dataset available for everyone to use for their research purposes. (C) 2020 Elsevier Ltd. All rights reserved.

Automated summary

The development of natural gas consumption forecasting tools is crucial for forecasting models, with research efforts focused on creating a new dataset with relevant data features. A forecasting methodology is proposed to evaluate statistical and machine learning algorithms in the time series forecasting domain, along with the availability of the new dataset for research use.