Analysis of EPC (Energy Performance Certificate) with the aim of identifying a model capable of synthesizing data and using it as training for deep learning models for the prediction and/or identification of energy consumption patterns for the building stock. Currently, data from the Lombardy region are being analyzed.
The repository is structured as follows:
- data: folder with different EPC dataset already processed
- models: DNN models and generative models
- EPC_analysis: generate NN models of EPC
- synthetic_model: generate synthetic model of EPC (filtred for labels A and A+)
- utils: functions to be used for the analysis
4 modles have been realized:
- A_EPC_NN_model: DNN model for EPC of class A and A+
- dnn_model: DNN model for all EPCs
- EPC_model_A: generative model to create synthetic EPC data of classes A and A+
- EPC_model: generative model to create syntehtic data of all EPCs