Big data analysis and prognosis of energy consumption and renovation costs in real estate
The heating of existing buildings contributes to a large extent to the energy consumption in Germany. About 40 % of the total energy is used to generate heating energy nationwide. The legislator has enacted regulations over several steps which are intended to minimise the energy consumption of new buildings and progressively also of existing buildings. Standardised calculation methods are used for this purpose. Although these methods can be used to make rough comparisons of the energy requirements of different residential buildings, the results generally deviate greatly from the actual energy consumption.
However, more accurate energy data are necessary for two main reasons: The promotion of energy-efficient construction is a sensible political goal in order to achieve climate protection goals and to reduce the costs of heat supply to an economically sensible level for present and future generations. Nevertheless, the associated methods for determining the (predicted) consumption must provide realistic values. Furthermore, neither planners, nor executing companies, nor the users of the residential buildings may be unsettled by partly misleading results and, on the basis of these results, may commission unsuitable and/or uneconomical structural or technical measures. In particular, it is important to avoid economically unreasonable effects.
As part of the BigDAPESI research project, a forecasting tool was developed for data-driven analysis and forecasting of energy requirements using machine learning algorithms. The algorithm used for the prognosis was trained and evaluated on the basis of approx. 15,000 anonymous input data of one- and two-family houses in Germany. The algorithm was specifically developed for a few but highly relevant attributes for an energy consumption forecast. So that a fast and reliable energy consumption forecast can be created.