4. GISUpdate2010 4
Jakarta
• Global City of the South - Fourth most
populous city in the world: 23 m
• Elevation 4m
• 87% Muslim
• Poverty, congestion, horrendous traffic,
stifling pollution, endemic corruption
6. GISUpdate2010 6
Focus
• The Image of The City by Kevin Lynch:
• the legibility of the city - "...the ease with
which its parts may be recognised and can be
organised into a coherent pattern."
• Our focus: the Desakota
• Longitudinal study: ‘a strip’
• Research questions..perception, opportunity,
impediment..
18. GISUpdate2010 18
Self Organising Map
• “a similarity graph, and a clustering diagram, too. Its
computation is a nonparametric, recursive regression
process” (Kohonen, 2000)
• Form of artificial neural network.
• Computationally intensive.
• unsupervised learning algorithm (neural networks)
19. GISUpdate2010 19
Purpose
• Represent high-dimensional data in a low-
dimensional form without loosing any of the
'essence' of the data.
• Organise data on the basis of similarity by
putting entities geometrically close to each
other.
31. GISUpdate2010 31
Principle Component Analysis
• Patterns among high dimensional data
• Calculates eigenvectors from covariance
matrices
• Measure variability in responses, discern
and account for groupings, identify
opposing groups
36. GISUpdate2010 36
Conclusion
• Exploring the Desakota
• Combination of EDA techniques
• ‘linking’ with video
• Further Develop methodology
• Apply in China and India – global cities of
the south..
Notes de l'éditeur
Indonesia
TOWERING GROWTH
Modern office blocks rise amid squalid neighborhoods lacking basic sanitation in the capital of the world's fourth most populous—and 87 percent Muslim—nation. Jakarta reflects Indonesia's vast resources, but also its astonishing problems, including: poverty, overpopulation (expected to increase 12-fold from 1950 to 2015), horrendous traffic, stifling pollution, and endemic corruption.
Jakarta, strip
earch revisits the work of urban designer Kevin Lynch, in particular his concepts of the ‘legible city’, ‘urban imageability’, and ‘cognitive mapping’. This proposal develops that preliminary investigation by drawing on three distinctive intellectual contexts: first, architecture, urban design and landscape theory; second, geography and Southeast Asian studies; and third, GIS science, public participatory GIS and geospatial hypermedia.
The project is sited in Jakarta, and draws on ethnographic research conducted by a group of researchers from the city. The research takes the form of video interviews designed to elicit the details of the interviewee’s everyday engagement with the city. This explores the routes they use, the territories they are familiar with, and their understanding of the city in general. The interview and video footage is enlarged upon by a mental map. These mental maps form the basis of our research, helping to develop a language appropriate to this city, and it’s desa-kota (peri-rural) condition which can resist traditional forms of mapping, drawing, and notation.
The significance of this research is three-fold. First, the desa-kota landscape itself deserves sustained academic attention as it represents an emergent mode of settlement that supports some of the largest population concentrations in the world. Research on this condition has the potential to contribute to (theoretical and policy) debates about the fortunes of this mode of settlement in Jakarta and Southeast Asia, and to link these debates to wider discussions on landscape urbanism, which are currently oriented almost exclusively towards European and American exemplars. Second, there has been, to date, little work on the ways in which visual media and representational systems – conventionally understood to be mutely instrumental – impact upon the design, planning and management of extended metropolitan regions.
RESEARCH QUESTIONS
1.How might international discussions on urban legibility, landscape urbanism, and the late capitalist city inform urban design and development in desa-kota zones?
2.What are the different visual systems by which desa-kota zones are represented and to what ends are they put: planning, property speculation, navigation, orientation?
3.How are desa-kota zones represented within the GIS-supported representational logics in use in the state urban and regional planning system?
4.What are the characteristics of the emergent, less formal representations of desa-kota zones that circulate in the popular media, NGO and community circles?
5.How do such representations relate to official representations: do they correspond, overlap, contradict, mix, or convolute?
6.What possibilities exist for the emergence of the less formal cultures of legibility in desa-kota zones, and for their interaction with official representations of desa-kota zones, without one subsuming the other?
7.How can new web-based, interactive geographic information technologies be used to explore these possibilities?
4.What are the characteristics of the emergent, less formal representations of desa-kota zones that circulate in the popular media, NGO and community circles?
5.How do such representations relate to official representations: do they correspond, overlap, contradict, mix, or convolute?
TOWERING GROWTH
Modern office blocks rise amid squalid neighborhoods lacking basic sanitation in the capital of the world's fourth most populous—and 87 percent Muslim—nation. Jakarta reflects Indonesia's vast resources, but also its astonishing problems, including: poverty, overpopulation (expected to increase 12-fold from 1950 to 2015), horrendous traffic, stifling pollution, and endemic corruption.
Source: http://www-vis.lbl.gov/Events/SC07/Drosophila/3DParallelCoordinates.png
View as a parallel coordinate plot, in which each plane represents a specific question, and each ‘thread’ through those planes represents a particular individual’s response.
BUT – what are the patterns in this data – are there meaningful ways by which we can group these responses? Are there outliers? Can we discern ‘new species’ or groups not previously identified in the Desakota?
artificial neural network
ANN - system loosely modelled on the human brain. An attempt to simulate the multiple layers of simple processing elements called neurons. Each neuron is linked to certain of its neighbours with varying coefficients of connectivity that represent the strengths of these connections. Learning is accomplished by adjusting these strengths to cause the overall network to output appropriate results.
Trial and error: uses process of learning, iterative nature. Similarity to biological neurons.
Input - set of n-dimensional observations. Output = network of nodes/akin to raster model with references to input data. Input vectors train the neuron grid so that topological relationships among input observations are preserved.
Complexity - formulas!
The SOM is a new, effective software tool for the visualization of high-dimensional data. It converts complex, nonlinear statistical relationships between high-dimensional data items into simple geometric relationships on a low-dimensional display. As it thereby compresses information while preserving the most important topological and metric relationships of the primary data items on the display, it may also be thought to produce some kind of abstractions. These two aspects, visualization and abstraction, can be utilized in a number of ways in complex tasks such as process analysis, machine perception, control, and communication.
The SOM algorithm computes the models so that they optimally describe the domain of (discrete or continuously distributed) observations.
Code junkies who want to make their own SOM
http://www.ai-junkie.com/ann/som/som1.html
World poverty map
http://www.cis.hut.fi/research/som-research/worldmap.html
Analaysis of questionnaires – generation of component planes
Ethnicity: Betawi, Javanese, Sudanese, Other
Betawi people - local inhabitant of Jakarta - descendants of the people living around Batavia (the colonial name for Jakarta)
km to doctor, km to school, km to bank, food and clothes
km to doctor, km to school, km to bank, food and clothes
Medium and large gated communities
Land tenure, length of time lived in community
Medium and large gated communities
Land tenure, length of time lived in community
Analaysis of questionnaires – generation of component planes
Principal component analysis (PCA)is used to find patterns amongst high dimensional data (Maindonald and Braun 2010; Shlens 2005). The technique involves calculating eigenvectors from the covariance matrix of the data. It is an exploratory data analysis technique that involves a mathematical procedure that transforms a number of possibly correlated variables into a smaller number of uncorrelated variables called principal components. PCA can be used to measure the variability in response among those questioned, help us to discern groupings, and how they are constituted. It can also help us to assess whether the groups are opposed. Maindonald, J and Braun, W.J. 2010 Data Analysis and Graphics Using R: An Example-based Approach. Cambridge University Press. Third edition. Shlens. J. 2005. A Tutorial on Principal Component Analysis. Copy retrieved [04-06-2010] from: http://www.cs.cmu.edu/~elaw/papers/pca.pdf But I need some words that describe in lay terms what the figure shows us: In your figure, are those variables furthest from the centre, those with the largest eigenvectors, and thus best able to account for the patterns discernible in the data? When we see opposed variables, (eg Desa and Kampung on one side, and Medium gated communities on the other), are we saying that 'those in gated communities are quite distinct from both the small rural village (Kampung) and the rapidly urbanising rural areas (Desakota)' Are we also saying that those in the Desakota are also a long way from various services (Banks, schools, shops)? Is this because the services have not yet arrived in the Desakota? What are we able to say about the proximity of 'long lived in the city and having Betawi ethnicity? I suppose what I need are some annotations of the PCA map that I can convey to the audience on Friday.
The output plane, cells sized according to the number of respondents in that group.
Jakarta, strip
earch revisits the work of urban designer Kevin Lynch, in particular his concepts of the ‘legible city’, ‘urban imageability’, and ‘cognitive mapping’. This proposal develops that preliminary investigation by drawing on three distinctive intellectual contexts: first, architecture, urban design and landscape theory; second, geography and Southeast Asian studies; and third, GIS science, public participatory GIS and geospatial hypermedia.
The project is sited in Jakarta, and draws on ethnographic research conducted by a group of researchers from the city. The research takes the form of video interviews designed to elicit the details of the interviewee’s everyday engagement with the city. This explores the routes they use, the territories they are familiar with, and their understanding of the city in general. The interview and video footage is enlarged upon by a mental map. These mental maps form the basis of our research, helping to develop a language appropriate to this city, and it’s desa-kota (peri-rural) condition which can resist traditional forms of mapping, drawing, and notation.
The significance of this research is three-fold. First, the desa-kota landscape itself deserves sustained academic attention as it represents an emergent mode of settlement that supports some of the largest population concentrations in the world. Research on this condition has the potential to contribute to (theoretical and policy) debates about the fortunes of this mode of settlement in Jakarta and Southeast Asia, and to link these debates to wider discussions on landscape urbanism, which are currently oriented almost exclusively towards European and American exemplars. Second, there has been, to date, little work on the ways in which visual media and representational systems – conventionally understood to be mutely instrumental – impact upon the design, planning and management of extended metropolitan regions.
RESEARCH QUESTIONS
1.How might international discussions on urban legibility, landscape urbanism, and the late capitalist city inform urban design and development in desa-kota zones?
2.What are the different visual systems by which desa-kota zones are represented and to what ends are they put: planning, property speculation, navigation, orientation?
3.How are desa-kota zones represented within the GIS-supported representational logics in use in the state urban and regional planning system?
4.What are the characteristics of the emergent, less formal representations of desa-kota zones that circulate in the popular media, NGO and community circles?
5.How do such representations relate to official representations: do they correspond, overlap, contradict, mix, or convolute?
6.What possibilities exist for the emergence of the less formal cultures of legibility in desa-kota zones, and for their interaction with official representations of desa-kota zones, without one subsuming the other?
7.How can new web-based, interactive geographic information technologies be used to explore these possibilities?
artificial neural network
ANN - system loosely modelled on the human brain. An attempt to simulate the multiple layers of simple processing elements called neurons. Each neuron is linked to certain of its neighbours with varying coefficients of connectivity that represent the strengths of these connections. Learning is accomplished by adjusting these strengths to cause the overall network to output appropriate results.
Trial and error: uses process of learning, iterative nature. Similarity to biological neurons.
Input - set of n-dimensional observations. Output = network of nodes/akin to raster model with references to input data. Input vectors train the neuron grid so that topological relationships among input observations are preserved.
Complexity - formulas!
The SOM is a new, effective software tool for the visualization of high-dimensional data. It converts complex, nonlinear statistical relationships between high-dimensional data items into simple geometric relationships on a low-dimensional display. As it thereby compresses information while preserving the most important topological and metric relationships of the primary data items on the display, it may also be thought to produce some kind of abstractions. These two aspects, visualization and abstraction, can be utilized in a number of ways in complex tasks such as process analysis, machine perception, control, and communication.
The SOM algorithm computes the models so that they optimally describe the domain of (discrete or continuously distributed) observations.