Thesis Implementación de una librería en python para generación automática de mapas urbanos
Date
2023
Authors
Journal Title
Journal ISSN
Volume Title
Program
Ingeniería Civil Informática
Departament
Campus
Campus Santiago San Joaquín
Abstract
Con el motivo de conocer qué sectores urbanos poseen indicadores de interés con valores homogéneos, de manera rápida y automática, se propone la implementación de una librería en Python con 5 algoritmos de clustering urbano distintos: Deep Modularity Networks (DMoN), Gaussian Mixture Model (GMM), K-Nearest Neighbours (KNN), Self Organized Map (SOM) y un método basado en la Teoría de la Información (TDI). También se ofrecen herramientas de comparación de mapas urbanos: Adjusted Rand Index (ARI) y Adjusted Mutual Information (AMI), y por último una documentación. Se desarrolla un caso de estudio con datos de la Región Metropolitana de Chile, se mide el uso de recursos de los algoritmos y se valida el correcto funcionamiento de estos al generar mapas urbanos de manera automática. También se valida el funcionamiento de las herramientas de comparación.
With the purpose of identifying urban sectors that exhibit homogeneous values of interest indicators quickly and automatically, the implementation of a Python library is proposed. This library includes 5 different urban clustering algorithms: Deep Modularity Networks (DMoN), Gaussian Mixture Model (GMM), K-Nearest Neighbours (KNN), Self Organized Map (SOM) and a method based on Information Theory (TDI). Also, urban map comparison tools are provided: Adjusted Rand Index (ARI) and Adjusted Mutual Information (AMI), and a documentation. A case study is conducted using data from the Metropolitan Region of Chile to measure the resource usage of these algorithms and validate their proper functioning in generating urban maps automatically. The functionality of the comparison tools is also validated.
With the purpose of identifying urban sectors that exhibit homogeneous values of interest indicators quickly and automatically, the implementation of a Python library is proposed. This library includes 5 different urban clustering algorithms: Deep Modularity Networks (DMoN), Gaussian Mixture Model (GMM), K-Nearest Neighbours (KNN), Self Organized Map (SOM) and a method based on Information Theory (TDI). Also, urban map comparison tools are provided: Adjusted Rand Index (ARI) and Adjusted Mutual Information (AMI), and a documentation. A case study is conducted using data from the Metropolitan Region of Chile to measure the resource usage of these algorithms and validate their proper functioning in generating urban maps automatically. The functionality of the comparison tools is also validated.
Description
Keywords
Clustering, Mapas, Urbanismo, Segregación
