Five models and ten predictors for energy costs on farms in the European Union

Abstract Energy costs are the main concerns of the agricultural stakeholders, because of their economic, environmental, and social impacts on the farms and the development of interrelated activities. In fact, it is important to save costs with the energy use to improve the profitability of the farms, but the level of these costs is often interlinked with the options to manage the energy consumption and the respective implications on sustainability. This framework highlights the importance of good management and planning for energy utilisation in the farming sector, namely to promote a balanced and integrated rural development. Considering these perspectives, this research intends to identify which factor, and how, impacted the energy costs in the European Union farms over the last decades. To achieve these objectives data from the Farm Accountancy Data Network database were considered for the European Union agricultural regions and the period 2013–2021. This statistical information was analysed through machine learning approaches following the procedures proposed by the software IBM SPSS Modeler. The linear support vector machine, regression, random forest, random trees, and the classification and regression tree are the most accurate models. On the contrary, the level of production, the size of farms, the economic and financial structure, and policy measures are the most important predictors. The findings here may be important insights for the European Union farming stakeholders, specifically to allow the design of policies for a more adjusted energy resources management.