Introduction
Energy demand is essential when planning for infrastructure and grid investment (TadahiroN., Shigeyuki H., 2010). In order to properly model the demand-side of the energy market,several approaches have been proposed as to the selection of the independent and explanatory variables or features, as they are called in the AI literature. The approaches are indeed vastand ever-expanding. In this paper, we develop linear regression models for the forecastingof nal energy and electricity for the industrial sector in six EU countries (Germany, theNetherlands, Sweden, Spain, Portugal, and Greece). The aim of the investigation is twofold:rst, to highlight the key drivers of these two types of energy consumption; and second, tointroduce the potential of what-if counterfactual analysis on the selected features.To this extent, various explanatory variables have been used and are discussed below asto their ability to inform our two types of industrial consumption.