LEE- Mimari Tasarımda Bilişim-Doktora

Bu koleksiyon için kalıcı URI

http://hdl.handle.net/11527/19396

Gözat

Konu "Bayesian belief networks" ile LEE- Mimari Tasarımda Bilişim-Doktora'a göz atma
  • Öge
    A decision support model based on Bayesian belief network to evaluate urban vibrancy
    (Graduate School, 2023-08-24) Bakraç Kırdar, Gülce ; Çağdaş, Gülen ; 523172004 ; Architectural Design Computing
    Urban liveability can be accepted as an umbrella term that covers all the factors make place a good neighborhood to live in. This thesis study recognized the versatility of urban liveability, and puts emphasis on vibrancy in the context of liveability. This research takes place value measures as reference to examine vibrancy. The selected categories are economic, image and use value of place. Economic place value draws from Carmona's (2019) compiled evidence, use value from Jacobs' (1961) diversity generators and Montgomery's (1998) indicators of successful urban places, and Nasar's (1998) likeability features for visual perception of urban places. Eminönü Central Business District (CBD) in Istanbul's Historical Peninsula serves as the study's testbed for exploring vibrancy-focused liveability. The thesis aims to present a data-driven decision support system to evaluate vibrancy-focused liveability dimensions. This study adopts a knowledge discovery process with Bayesian Belief Network (BBN) to handle the complexity of the liveability concept. The thesis study questions how are the relationship patterns between urban vibrancy measures, which parameters can be prioritized based on this relationship network, and what kind of implications can be made regarding the urban vibrancy of the site. The objectives of the study are to develop a comprehensive measurement technique using multiple data types for the measurement of urban vibrancy; to reveal the relationship network of vibrancy parameters; and to improve the decision-making process according to relationship network. The research hypothesis posits that the big data supported knowledge discovery process can be useful to reveal complex urban dynamics, and support vibrancy decisions together with participation. Methodologically, this study adopts quantitative research. This study presents exploratory research through the use of big data and BBN analysis to examine the vibrancy focused liveability with spatial, functional and perceptual attributes. The thesis study explores the relationship through BBN and explores causality through the consultation of expert opinion. A causal knowledge discovery process involves data collection, information retrieval, and data analysis. Data collection involves techniques like web scraping and urban map digitization; while information retrieval encompassing quantitative methods which are entropy-based indices, clustering algorithms, image segmentation, and surveys. Data analysis employs BBN learning algorithm to unveil probabilistic relationships between place value measures, and calibration of BBN network with expert participation via surveys. Spatial distribution results and BBN analyses provide insights into vibrancy levels and priority measures to enhance place value. The results demonstrate that urban function and accessibility outweigh urban form and socio-demographic features in determining economic value (land price). Activity characteristics and heritage within accessibility enhance use value (user density), while nature and cultural elements positively impact image value (likeability), countered by negative influences from signboards and building enclosures. Economic value BBN reveals that land use diversity has the most substantial impact on land price, followed by building density, other land use characteristics, accessibility, and urban form features. The use value BBN model highlights the significance of heritage visitation, density, and activity accessibility in hotspot user density, followed by activity diversity, density, and distribution. The image value BBN model indicates that increased urban greening, vistas, and cultural landscapes enhance likeability, while building enclosures and façade signboards have negative effects. Tahtakale, Beyazıt, Eminönü, Sirkeci, and Sultanahmet are highly vibrant districts, and Hobyar, Rüstempaşa, Alemdar, Binbirdirek, Sultanahmet, and Beyazıt are highly vibrant neighborhoods. In the survey, expert participants rank place values, determine causality and correlation of relationships between parameters. Correlations between BBN and survey data validate the creation of a causal map. The correlation between BBN and survey data confirms that survey data can be used to create a causal map. Regarding the causal relationships, prioritizing urban function and accessibility measures in economic value metrics will aid in developing real estate strategies. To enhance use value, the activity diversity and accessibility, attractiveness and visitation of heritage, can be prioritized, which contribute on place attractiveness decisions. To improve image value, urban greening, landscape and building façade, and signboard density measures can be prioritized, which contribute on maintenance decisions of the streetscape. The decision support system (DSS) contributions to urban planning and design have been assessed with what-if analysis using spatial BBN tools and urban design workshop. This data-driven approach supports conceptual decisions in urban design, and prioritizes decisions in urban planning. This research aims to assist decision-makers in creating vibrant neighborhoods through data-driven methods. This study would be useful for urban planners to generate inclusive spatial strategies by considering human activity factors within physical attributes to create vibrant neighborhoods.