An Approach to Estimation of Residential Housing Type Based on the Analysis of Parked Cars

摘要

A method for prediction of residential housing types based on an analysis of the number of cars parked near buildings in consideration is proposed in the paper. The source of data constitute satellite or aerial images of a given residential area where cars and building can be identified. The machine learning models are build based on the distribution of car parked in the area. The resulting classification models allow for distinguishing between low-rise, mid-rise and high-rise housing. The effectiveness of the method was proved using aerial images of three residential districts of a big city in Poland and the WEKA data mining system.