Abstract
Reviewed by: Ron Johnston, University of Bristol, UK DOI: 10.1177/2399808318759193
Geocoded big data sets visualised and analysed within software packages are at the core of a newly named academic activity – urban analytics. According to a leader in the field profiled in this book, big data are changing how ‘business is conducted, governance enacted, and science undertaken’: we get some feel for the last of these, but little on the other two.
Urban analytics is short and heavily illustrated, designed to help students and professionals utilise those data. The authors claim it differs from other texts which eschew quantitative techniques whereas Urban analytics exposes readers to ‘a novel set of tools, techniques and data’, although the authors ‘confess’ the material is ‘pitched at a high level’.
The first chapter introduces big data, stressing not only volume but also velocity and variety. They are also fine-grained in resolution – usually collected from individuals with some form of geocoding – and readily merged with other data sets, and they are exhaustive, ‘striving to capture entire populations or systems … or at least much larger sample sizes than would be employed in traditional, small data studies’. Not stressed, however, is that many of these data sets are collected by or for commercial enterprises and do not correspond to the carefully selected samples that characterize other types of survey with which ‘conventional’ statistical procedures can be deployed. The data are presented as unproblematic; are they?
Three further chapters introduce: the ways in which data are collected; the infrastructure within which that, and subsequent visualisation and analysis, occurs (in effect, the chapter is a primer on the nature of computers); and how the city can be visualised – ‘effective representation’ is the chosen term – using computer software. Here is one of the book’s biggest failings because it frequently fails to deliver effective representation. There are many colour figures, including maps, of which a substantial number, in part or in whole, are illegible – in particular the scales and keys are too small to be read. Furthermore, these maps are presented as little more than trophies with very little, if any, interpretation.
Interpretation is even more important in later chapters. One on ‘Differences within cities’ concentrates on geodemographic classifications but with little detail on the illegible figures and nothing on their use. A chapter on ‘Explaining the city’ concerns analysis – discerning patterns within noise – with spatial statistics briefly introduced. There is only one worked example, a regression of the per cent with health insurance in subdivisions of Phoenix (although the analysis is of proportions): the regression coefficients are reported to five decimal points with no reference to the relevance of significance levels in analyses of very big, possibly unrepresentative, data sets. A chapter on ‘Generative urban systems’ deals with modelling their structures and the behaviour within them – which is at a high level with terminology that may not be readily assimilated. The examples are not described: one supposedly shows different traffic volumes along roads in Quinto [sic], Ecuador; different segments are red and green, but there is no indication of what they represent. The last substantive chapter is on ‘Cities as networks and flows’; one example is a map of surname regions in Great Britain created by clustering ‘using Lasker distances’ (which Wikipedia – but not this book – tells me are the ‘negative value of the logarithm of isonymy between localities’).
A final chapter on ‘The future of urban analytics’ claims the book has aimed ‘to consolidate progress across urban planning, design, geography, and the information sciences, but particularly to emphasize the role of geographic information science [a term not in the index] for tackling grand urban challenges’. Associated with the book are two sets of online resources: a set of PowerPoint slides that lecturers can use; and practical labs for students – which after some time exploring I decided were user-unfriendly.
This book represents a lost opportunity – despite the quotes from promoters on the back cover. There is great potential in visualising and analysing the great volume of data now being made available, which may be useful for tackling grand urban challenges (such as?). Having described their nature and the infrastructure – in more and better illustrated detail – the authors would have been better served if they had taken Rob Kitchin’s advice, cited on page 25: ‘try and start with a question and then find the data to try and answer that, rather than starting with the data and trying to find something useful to do with them’. The current book suggests that urban analytics is little more than bland, if colourful, empiricism.