ML_for_usable_area_estimation_gable_roofs

Machine Learning of Usable Area of Gable-Roof Residential Buildings Based On Topographic Data

This repository contains all the necessary data and code to reproduce results and figures from paper "Machine Learning of Usable Area of Gable-Roof Residential Buildings Based On Topographic Data" by L. Dawid, K. Cybiński, and Ż. Stręk, namely:

• Folder Gable-roof contains the datasets we used to train, and test our Machine Learning (ML) models
• Folder models contains the models we chose as performing the best, which's performance has been presented in the paper on Figs 5-7
• Jupyter notebook Direct_comparison.ipynb, which has a section including model training, and comparison of Linear Regression, and Neural Network (NN) performance on different dataset combinations
• auxiliary_funcs.py is a technical file containing NN and Linear Regression training routines, as well as other function definitions necessary to reproduce Figs 5-7
• Jupyter Notebook MC.ipynb is dedicated to reproducing the LiDAR height detection error estimation, and reproduction of Figs 2-4

Code has been written by Kacper Cybiński (University of Warsaw)