Machine Learning - Typology
Classifying Architectural Typologies with Unsupervised Learning
Independent — 2017

In Progress...

This research, currently in progress, applies unsupervised learning techniques to the classification of building typology. Using python code and physical data from a New York City’s Department of City Planning dataset, this algorithm takes an address as an input, analyzes the site’s dimensions and proportions, and outputs a prediction of building type (i.e. single family residence, warehouse, cathedral). 

Applied techniques include data exploration, benchmarking, data preprocessing, feature transformation, Gaussian mixture
model clustering, and k-means clustering.


Copyright © 2020 Jim Peraino.
All rights reserved.