A carregar...
Publicação
Mobilidade Intermunicipal em Transporte Público na Área Metropolitana de Lisboa: uma abordagem espacial com modelos geográficos de floresta aleatória e regressão ponderada multiescala
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 7.21 MB | Adobe PDF |
Orientador(es)
Resumo(s)
A mobilidade intermunicipal em transporte público coletivo (TC) nas áreas metropolitanas é um
elemento fundamental do planeamento urbano, especialmente na promoção de sistemas de
transporte sustentáveis e equitativos. Este estudo foca-se na Área Metropolitana de Lisboa
(AML), através da análise dos fatores que influenciam os movimentos pendulares (movimentos
entre o local de residência e o local de trabalho/estudo) intermunicipais em TC (variável
dependente) ao nível da freguesia. A análise baseia-se em dados dos Censos de Portugal de
2021, incluindo atributos de mobilidade, infraestrutura de transportes e acessibilidade,
condições socioeconómicas e características do ambiente construído. Foram aplicados dois
modelos, sendo estes o Python-based Geographical Random Forest (PyGRF) e o Multiscale
Geographically Weighted Regression (MGWR). O PyGRF, um modelo de machine learning
espacialmente explícito, destacou-se pelo desempenho preditivo superior na identificação de
variáveis explicativas relevantes assim como a visualização das características locais. O MGWR,
baseado em pressupostos lineares, complementou a análise com uma abordagem multiescala,
ao revelar coeficientes locais estatisticamente significativos. A integração destes modelos,
através de mapas coropléticos bivariados, permitiu uma visão abrangente dos principais fatores
que influenciam os movimentos pendulares em TC. Os resultados evidenciam a necessidade de
políticas direcionadas à melhoria das infraestruturas de TC, sobretudo na periferia, com ênfase
nas freguesias da margem sul do rio Tejo. Este estudo fornece evidências robustas para apoiar
políticas urbanas que promovam uma mobilidade sustentável e equitativa.
Intermunicipal mobility using public transport in metropolitan areas is a crucial element of urban planning, particularly in promoting sustainable and equitable transportation systems. This study focuses on the Lisbon Metropolitan Area (LMA) by analyzing the factors influencing intermunicipal commuting (movements between place of residence and place of work/study) using public transport (dependent variable) at the parish level. The analysis is based on data from the 2021 Portugal Census, including mobility attributes, transport infrastructure and accessibility, socioeconomic conditions, and built environment characteristics. Two models were applied: the Python-based Geographical Random Forest (PyGRF) and the Multiscale Geographically Weighted Regression (MGWR). The PyGRF, a spatially explicit machine learning model, stood out for its superior predictive performance in identifying relevant predictors and visualizing local characteristics. The MGWR, grounded in linear assumptions, complemented the analysis with a multiscale approach by revealing statistically significant local coefficients. The integration of these models through bivariate choropleth maps provided a comprehensive view of the main factors influencing public transport commuting patterns. The results highlight the need for targeted policies to improve public transport infrastructure, particularly in peripheral areas, with a focus on the municipalities on the southern bank of the Tagus River. This study provides robust evidence to support urban policies that promote sustainable and equitable mobility.
Intermunicipal mobility using public transport in metropolitan areas is a crucial element of urban planning, particularly in promoting sustainable and equitable transportation systems. This study focuses on the Lisbon Metropolitan Area (LMA) by analyzing the factors influencing intermunicipal commuting (movements between place of residence and place of work/study) using public transport (dependent variable) at the parish level. The analysis is based on data from the 2021 Portugal Census, including mobility attributes, transport infrastructure and accessibility, socioeconomic conditions, and built environment characteristics. Two models were applied: the Python-based Geographical Random Forest (PyGRF) and the Multiscale Geographically Weighted Regression (MGWR). The PyGRF, a spatially explicit machine learning model, stood out for its superior predictive performance in identifying relevant predictors and visualizing local characteristics. The MGWR, grounded in linear assumptions, complemented the analysis with a multiscale approach by revealing statistically significant local coefficients. The integration of these models through bivariate choropleth maps provided a comprehensive view of the main factors influencing public transport commuting patterns. The results highlight the need for targeted policies to improve public transport infrastructure, particularly in peripheral areas, with a focus on the municipalities on the southern bank of the Tagus River. This study provides robust evidence to support urban policies that promote sustainable and equitable mobility.
Descrição
Dissertation presented as the partial requirement for obtaining a Master's degree in Geographic Information Systems and Science, specialization in Geospatial Data Science
Palavras-chave
Movimentos pendulares intermunicipais Transporte público coletivo Centralidade metropolitana Geographical Random Forest Multiscale Geographically Weighted Regression Intercity commuting movements Public transportation SDG 11 - Cidades e comunidades sustentáveis