**6. Conclusions**

The excessive demand for energy is a major challenge for countries. Therefore, governments seek to improve energy managemen<sup>t</sup> and efficiency to reduce energy waste. In this article, a hybrid approach based on clustering and classification proposes discovering factors affecting energy efficiency in the domestic sector.

49,815 examples of the housing stock of England and Wales were used. First, households were analyzed to identify the influence of factors using a decision tree (without using the proposed approach). Then, the proposed approach was used. The K-means algorithm yields three clusters (low-consumption, medium-consumption, and high-consumption clusters). Households in each cluster were analyzed using the C5.0 algorithm. Comparing the results and modeling accuracy, once without using the approach and then using it, showed the ability of the approach presented to identify the properties that affect energy efficiency and consumption in-depth and more accurately. The approach presented is adaptable to different data sets.

The results of not using the approach shown is that installing boilers has improved the energy efficiency, especially with respect to those that dwelling installed in 2009, followed by boilers installed in 2005 and 2012. Dwelling install boilers 2010 have the least performance. Using electricity tariff 7 results in poor efficiency. Also, in old homes the thermal equipment and energy consumption of the equipment are weak compared to newly-built houses, which results in in weaker energy efficiency. The data also showed that walls which have a cavity structure as well as insulating installation lead to improved energy efficiency. Of course, the cavity wall itself has different types, which produces comments on how these walls affect the energy efficiency. The need for information on the type of cavity wall used in the residential buildings is urgen<sup>t</sup> to find out more thorough knowledge.

Besides the findings presented above, the proposed approach provides new and more detailed results. It is demonstrated that electricity tariff 7 has different behaviors in different clusters. Generally, it was seen that the use of this tariff is not good to improve energy efficiency. The approach shows

that in the low-consumption cluster, old and small houses (less than 51 m2) that use this tariff have a poor energy efficiency. Also, among the old and big houses (over 151 m2) of the medium-consumption cluster using this tariff has a positive impact.

The approach shows that the home structure influences energy efficiency. In the high-consumption cluster, installing boilers in mid-terrace and end-terrace structures, and detached structures in Wales in the medium-consumption cluster, leads to better energy efficiency. In the medium-consumption cluster, it was seen that flats are better in energy efficiency than other structures, even compared to newly-built houses.

Different geographic regions also had a different behavior. The high-consumption cluster shows better energy efficiency in the houses in Wales. Definitely, having more comprehensive and adequate information of the different regions of England and Wales could extract more knowledge. The accuracy of modeling in the approach presented was better than modeling without it and detailed findings can be discovered.

Through its new and in-depth results, this approach has shown that it is capable and beneficial in the field of retrieval knowledge. These new findings demonstrate that we cannot make a similar decision for all homes. As homes in a cluster have their unique behavior, policies and decisions must be unique for them. The knowledge obtained is suitable and useful for residential buildings of similar features and nature to plan and upgrade energy efficiency, and also to improve EPCs.

**Author Contributions:** M.N. and A.H. proposed the idea; M.N. designed the model and the computational framework, also carried out the implementation and processed the experimental data, performed the analysis and designed the figures, interpreted the results and wrote the manuscript, performed the proofreading, discussed the results; A.H. verified the analytical methods and supervised the findings of this work, discussed the results; F.M.-Á. verified the analytical methods, discussed the proofreading, contributed in discussing the results. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors thank the reviewers for their valuable suggestions for improving the manuscript. This research did not receive any specific gran<sup>t</sup> from funding agencies in the public, commercial, or not-for-profit sectors.

**Conflicts of Interest:** The authors declare no conflict of interest.
