Rozdziały

Stały URI dla kolekcjihttp://hdl.handle.net/11652/5022

Przeglądaj

Filtry

20232

Ustawienia

Tematsearch.filters.applied.operator.equals: search.filters.subject.machine learning

collection.search.results.head

Teraz wyświetlane 1 - 2 z 2
  • Pozycja
    On the relationship between urban form and amenities: A new perspective from Qom (Iran)
    (Lodz University of Technology Press, 2023) Venerandi Alessandro; Zamani Vahid; Porta Sergio
    Amenities are fundamental for urban life as they promote socio-economic interactions and enhance city dynamics. Previous studies investigated the relationship between metrics of street network centrality and urban amenities. However, they hardly focused on further aspects of the built environment. A further drawback is that relationships were mainly assessed through linear models even though more complex and non-linear relationships plausibly exist. In this work, we, first, comprehensively describe the urban form of our case study, the city of Qom (Iran), through a set of 55 morphometrics computed at the plot level; second, we investigate the relationship between these metrics and density of amenities, through a set of machine learning techniques that handle non-linear behaviours. The best model explains up to 45% of the variance of the density measure, with coverage ratio, plot size, floor area ratio, street canyon width, and betweenness centrality being the top five explanatory factors. While the findings of this work do not have universal value, the methodology can be replicated to explore the same research question in different contexts. It can also be used as an evidence-based tool to inform design choices in urban redevelopment affecting the location of amenities in cities.
  • Pozycja
    Comparison analysis on typical historic cultural districts with AI machine learning technology – Taking Portuguese and Macao districts as examples
    (Lodz University of Technology Press, 2023) Jiang Shan; Zheng Liang; Chen Yile; Zheng Jianyi
    With the rapid development of technology, artificial intelligence has gone into every field, and its development has been further expanded with machine learning as the core technology. How does this help urban analysis and urban form research? This study aims to introduce a new method for analysing and comparing urban morphological layouts using machine learning technology and to explore the possibility and potential of combining urban morphology analysis with machine learning technology. In this exploratory study, several typical Portuguese cities with historical and cultural characteristics are used as learning samples for comparison. Through the combination of urban morphology theory and machine learning, the urban morphological samples are clipped out from the Portuguese city maps, then morphological features are extracted from the samples, establishing training labels as typical Portuguese urban fabric, lastly compared the result with the typical urban areas of Macao using the YOLOv4 object detection algorithm. Through the research, it is found that Macao in the early stage is more morphologically similar to the city of Evora due to their same privilege; after the early 20th century, influence by contemporary Portuguese engineers and urban development strategy, Macao's urban morphology shows a higher degree of similarity to that of Lisbon.