Developing ML Models for Early Stage Building Energy Prediction (using CNN)
Description
The ML model for early stage building energy predictions is developed using convolutional neural network (CNN) approach, proposed by [1] M.M. Singh, P. Geyer, CNN-based Quick Energy Prediction Model using Image Analysis for Shape Information, in: 17th Int. IBPSA Build. Simul. Conf., Brugges, Belgium, 2021. https://www.researchgate.net/publication/354420055
Files
Steps to reproduce
Use python to run Program/RunCNNAnalysis.py to train ML models. It uses the training samples from TrainData/Building.csv files. TrainData/BuildingShapeData.csv contains image representation of building footprint for each sample. It will use TestData folder to make prediction on the test dataset and evaluate the ML model. Change "MunichDEU" to "BerlinDEU" or "BrusselsBEL" for other locations.