Energy efficiency and environmental performance optimization at the district level are following an upward trend mostly triggered by minimizing the Global Warming Potential (GWP) to 20% by 2020 and 40% by 2030 settled by the European Union (EU) compared with 1990 levels. This paper advances over the state of the art by proposing two novel multi-objective algorithms, named Non-dominated Sorting Genetic Algorithm (NSGA-II) and Multi-Objective Harmony Search (MOHS), aimed at achieving cost-effective energy refurbishment scenarios and allowing at district level the decision-making procedure. This challenge is not trivial since the optimisation process must provide feasible solutions for a simultaneous environmental and economic assessment at district scale taking into consideration highly demanding real-based constraints regarding district and buildings’ specific requirements. Consequently, in this paper, a two-stage optimization methodology is proposed in order to reduce the energy demand and fossil fuel consumption with an affordable investment cost at building level and minimize the total payback time while minimizing the GWP at district level. Aimed at demonstrating the effectiveness of the proposed two-stage multi-objective approaches, this work presents simulation results at two real district case studies in Donostia-San Sebastian (Spain) for which up to a 30% of reduction of GWP at district level is obtained for a Payback Time (PT) of 2–3 years.

Artificial Neural Networks (ANNs) are weighted directed graphs of interconnected neurons widely employed to model complex problems. However, the selection of the optimal ANN architecture and its training parameters is not enough to obtain reliable models. The data preprocessing stage is fundamental to improve the model’s performance. Specifically, Feature Normalisation (FN) is commonly utilised to remove the features’ magnitude aiming at equalising the features’ contribution to the model training. Nevertheless, this work demonstrates that the FN method selection affects the model performance. Also, it is well-known that ANNs are commonly considered a “black box” due to their lack of interpretability. In this sense, several works aim to analyse the features’ contribution to the network for estimating the output. However, these methods, specifically those based on network’s weights, like Garson’s or Yoon’s methods, do not consider preprocessing factors, such as dispersion factors , previously employed to transform the input data. This work proposes a new features’ relevance analysis method that includes the dispersion factors into the weight matrix analysis methods to infer each feature’s actual contribution to the network output more precisely. Besides, in this work, the Proportional Dispersion Weights (PWD) are proposed as explanatory factors of similarity between models’ performance results. The conclusions from this work improve the understanding of the features’ contribution to the model that enhances the feature selection strategy, which is fundamental for reliably modelling a given problem.