Training and Test Dataset for Early Stage Energy Prediction Model using Advanced Machine Learning Approaches

The dataset is related to the article titled "Convolutional Neural Network to Learn Building Shape Representations for Early Stage Energy Prediction". It contains simulation data from EnergyPlus that is processed in a structure to train machine learning model using approaches of artificial neural network (ANN), component-based machine learning (CBML) model, and convolutional neural network (CNN) model. The dataset contains raw data, programs to reproduce the data, trained machine learning (ML) models, and their predictions. It also contains an auto encoder model used to reduce dimensionality of raw three-dimensional input to the CNN model.本数据集关联于题为《面向建筑早期能耗预测的建筑形态表征学习卷积神经网络》的研究论文。其包含来自EnergyPlus能耗模拟软件的仿真数据，经结构化处理后，可用于训练采用人工神经网络（Artificial Neural Network, ANN）、基于组件的机器学习（Component-Based Machine Learning, CBML）以及卷积神经网络（Convolutional Neural Network, CNN）方法的机器学习模型。本数据集涵盖原始数据、复现该数据集的程序、已训练完成的机器学习（Machine Learning, ML）模型及其预测结果。此外，数据集还包含用于将三维原始输入降维后输入至CNN模型的自编码器（Autoencoder）。