AUM2020 Global Workshop: Session 8: Modelling method (2)
Duration: 2 hours
Share this media item:
Embed this media item:
Embed this media item:
About this item
Description: |
Session outline:
1. Synthetic population generation using GANS and expert knowledge Mr Gael Lederrey, Dr Tim Hillel, and Prof. Michel Bierlaire (École polytechnique fédérale de Lausanne (EPFL), Switzerland) 2. High-resolution air temperature mapping in data-scarce areas by means of low-cost mobile measurements and machine learning Mr Ahmed H. M. Eldesoky (Università Iuav di Venezia), Prof. Nicola Colaninno (Politecnico di Milano) and Dr Eugenio Morello (Politecnico di Milano) 3. Building and validating modular urban transportation models using scientific workflow systems Dr Juste Raimbault and Prof. Michael Batty (University College London) Discussant: Dr Ying Jin (University of Cambridge) Host: Dr Kaveh Jahanshahi (University of Cambridge) |
---|
Created: | 2021-02-08 21:11 |
---|---|
Collection: | Martin Centre AUM2020: Modelling the New Urban World |
Publisher: | University of Cambridge |
Copyright: | The Martin Centre |
Language: | eng (English) |
Keywords: | AUM2020; Modelling the New Urban World; Martin Centre; Online Global Workshop; Architecture; |
Abstract: | 1. Agent-based simulations used for land-use and transportation modelling rely on accurate virtual representations of
the population of interest. They typically make use of synthetic populations generated from sample data which is assumed to represent the population of interest. In this work, we present a new methodology for synthetic population generation, which allows expert knowledge to be combined with powerful, flexible deep learning methods. Our approach, called the Directed Acyclic Tabular GAN (DATGAN), uses a Directed Acyclic Graph (DAG) to allow the modeller to specify complex dependencies between synthesised variables in the population. 2. Understanding urban micro- and local climates requires the availability of air temperature information at an effective spatial and temporal resolution. This study aims at providing such information to inform urban design and planning practices in a data-scarce, arid area. The objective is to produce accurate air temperature maps at a high spatial resolution for an entire city using air temperature data, collected from low-cost mobile measurements; different spectral indices, retrieved from freely-available satellite imagery; spatial analysis techniques; and random forest regression models. Thereafter, we explore the spatial variability of air temperature and quantify the urban heat island intensity. 3. Large scale urban transportation models such as four-step models require the integration of heterogenous data and the coupling of sub-models which can already be consequent in terms of complexity. Therefore, such integrated models are difficult to transfer, reproduce, and validate. We propose a modular and reproducible approach based on scientific workflow systems to build and validate such models. We illustrate it by coupling different open-source components within workflows to construct a four-step transportation model applied to all functional urban areas in the UK, and discuss its application to health indicators within public transport in the context of the COVID-19 crisis. |
---|
Available Formats
Format | Quality | Bitrate | Size | |||
---|---|---|---|---|---|---|
MPEG-4 Video | 640x360 | 990.05 kbits/sec | 870.17 MB | View | ||
WebM | 640x360 | 502.23 kbits/sec | 441.42 MB | View | ||
iPod Video | 480x360 | 488.06 kbits/sec | 428.97 MB | View | ||
MP3 | 44100 Hz | 250.61 kbits/sec | 220.26 MB | Listen | ||
Auto * | (Allows browser to choose a format it supports) |