Functional regions within the city centre: a study by factor analysis of taxi flows in central london
TL;DR: In this article, taxi flows are used as an indicator of multi-faceted functional linkages within Central London, and correlation and factor analysis are applied to the data to determine the underlying structure of the taxi-flow system.
Abstract: This paper examines the problem of measuring the relationship between movement patterns and the location of activities within the city centre. Emphasis is placed on the problem of defining and analysing complex linkage systems. Data on taxi flows are presented as a single indicator of multi-faceted functional linkages within Central London. Cartographic analysis reveals a complex pattern of flows; correlation and factor analysis are therefore applied to the data to determine the underlying structure of the taxi-flow system. Traffic zones are compared in terms of the zones from which they receive trips; high factor loadings reveal groups of zones receiving trips from common origins; the related factor scores specify these origins. Linking groups of origins and destinations on a map reveals five functional regions. Hierarchical grouping of zones produces an objective classification of zones in terms of similarities in linkage characteristics. Comparison of the pattern of functional regions with familiar structural regions suggests a basic isomorphism between such regionalizations. More important, the analysis reveals the usefulness of the methods for investigating the structure of complex flow systems so as to specify the principal groups of observations and the linkages they have in common. THE CLOSE connection between movement and the distribution of activities in spatial systems is a well-established geographical axiom. At the national level B. J. L. Berry has used multivariate statistical procedures to measure the relationships between the state of a system, expressed in terms of the distribution of productive activities, and the behaviour of that system, expressed in the form of commodity flows between various places.1 He has structured these relationships within the framework of a general field theory of spatial behaviour, in which the pattern of flows, summarized as a set of functional regions, is held to be both a resultant and a determinant of the spatial distribution of activities, summarized as a set of uniform regions.2 Berry's formulation is in some ways a more general statement of A. K. Philbrick's principle of'areal functional organization'.3 Similarly, O. Wameryd, in his conceptual model of the urban system, has stressed the importance of functional interdependencies in explaining changes in the state of the system.4 It would be reasonable to assume that the interdependence of flows and spatial structure also holds within urban areas, in particular within large metropolitan centres. However, at this level the measurement of the relationships involved creates numerous conceptual, empirical and technical problems, many of which are insuperable at the present time. While several studies of city centre structure have emphasized the role of 'linkages' between highly specialized activities as a determinant of location, little progress has been made in measuring these linkages. 5 Most rrevious research has focused on the distribution of activities; linkages have then been inferred from locational associations among these activities.6 For instance, within the financial core of the City of London, multivariate analysis of detailed employment data has highlighted the existence of distinctive spatial clusters of offices whose composition closely reflects the wellknown functional organization of the City.7 Within the broader compass of the London Central Area, floorspace data have suggested a number of sub-areas with distinctive land-use combinations.8 I6I This content downloaded from 157.55.39.35 on Wed, 31 Aug 2016 04:35:39 UTC All use subject to http://about.jstor.org/terms According to Berry's general field theory formulation, such pronounced spatial differentiation within city centres should simultaneously be determined by, and call forth, a correspondingly structured pattern of movement. Yet very few studies have been made of physical linkage patterns in the city centre, in relation to the activities that generate and attract these movements. In part this is because of data difficulties. While detailed information can be collected on the distribution of activities, a corresponding breakdown of movement patterns associated with each of these activities would require extensive surveys. More important, linkages frequently involve information flows that are extremely difficult to differentiate, that use a variety of communication channels and which often do not involve obvious physical movements.9 Some of these difficulties can be by-passed by considering only physical movements of a given type, as a single indicator of multi-faceted functional linkages. But even then, observed movement patterns are extremely complex. It is therefore the object of this paper to suggest a method for analysing the complex linkage structures that are represented by movement patterns in the city centre. As the first stage in the scientific description of any complex pattern involves the development of a classificatory framework, numerical taxonomic procedures are applied to observed patterns of flows. This yields a more readily appreciated series of functional regions that can be compared with uniform regions derived from a similar analysis of activity distributions. The Data and Study Area For the purposes of this paper, the movements of taxis within Central London have been taken as the best available single indicator of functional linkages. Of the principal travel modes, the use of taxis is most characteristic of the Central Area; in fact, 55 per cent of all taxi trips in Greater London both begin and end in the centre. 10 However, taxis still represent only one type of communication and the subsequent analysis is therefore forced to ignore linkages maintained on foot, by private vehicles, public transport, telephone and post. The Registrar General's definition of the Central Area for the I96I census is taken as the limit for the study area, which for practical purposes has to be regarded as a closed system (Fig. i). This area was subjectively extracted from the existing administrative framework by aggregating enumeration districts to provide an additional unit for collecting future census and planning data. Defined in this way, the Central Area includes an area of 28 km2, which is much larger than the Central Business District defined using R. E. Murphy and J. A. Vance's criteria.11 In fact, the Central Area includes both C.B.D. 'core' and 'frame' using E. M. Horwood and R. R. Boyce's terminology, thus containing the main-line rail termini, which are major generators of taxi traffic. 12 The data were collected in I962 as part of the London Traffic Survey. They are based on journey logs kept by an effective 10 per cent sample of all cabs registered in the Survey Area for one week in July.13 As the logs were completed by taxi drivers, the data failed to differentiate trips according to passenger's journey purpose. An additional shortcoming of the data from the linkage standpoint is that non-fare paying trips (that is, trips made between setting down one passenger and picking up another) are not excluded. Because drivers tend to cruise back into the areas where most trips are generated, the major nodes in the system could be over-emphasized. Both of these limitations are overcome in a second survey of taxi usage conducted in July I969 as part of a Home Office investigation into the London taxicab trade. Trip origins and destinations determined from the logs have been assigned to seventy I62 J. B. GODDARD This content downloaded from 157.55.39.35 on Wed, 31 Aug 2016 04:35:39 UTC All use subject to http://about.jstor.org/terms FUNCTIONAL REGIONS IN CENTRAL LONDON FIGURE I-London Central Area FIGURE 2-Traffic zones traffic zones in the Central Area and averaged to give 24-hour weekday flows. All but eight of the zones consist of aggregations of enumeration districts delineated with due consideration of land-use characteristics. The remaining eight zones comprise sections of major shopping streets, for example, Oxford Street (three zones) and Piccadilly (two zones) (Fig. 2). i63 This content downloaded from 157.55.39.35 on Wed, 31 Aug 2016 04:35:39 UTC All use subject to http://about.jstor.org/terms
Citations
More filters
TL;DR: In this article, the authors use the term social sensing for individual-level big geospatial data and the associat- tation of the data to understand the socioeconomic environments.
Abstract: The emergence of big data brings new opportunities for us to understand our socioeconomic environments. We use the term social sensing for such individual-level big geospatial data and the associat...
560 citations
Cites methods from "Functional regions within the city ..."
...Additionally, data sources such as taxi data were used to measure spatial interactions as early as 1970 (Goddard 1970)....
[...]
TL;DR: This paper introduces the concept of latent activity trajectory (LAT), which captures socioeconomic activities conducted by citizens at different locations in a chronological order, and develops a topic-modeling-based approach to cluster the segmented regions into functional zones leveraging mobility and location semantics mined from LAT.
Abstract: The step of urbanization and modern civilization fosters different functional zones in a city, such as residential areas, business districts, and educational areas. In a metropolis, people commute between these functional zones every day to engage in different socioeconomic activities, e.g., working, shopping, and entertaining. In this paper, we propose a data-driven framework to discover functional zones in a city. Specifically, we introduce the concept of latent activity trajectory (LAT), which captures socioeconomic activities conducted by citizens at different locations in a chronological order. Later, we segment an urban area into disjointed regions according to major roads, such as highways and urban expressways. We have developed a topic-modeling-based approach to cluster the segmented regions into functional zones leveraging mobility and location semantics mined from LAT. Furthermore, we identify the intensity of each functional zone using Kernel Density Estimation. Extensive experiments are conducted with several urban scale datasets to show that the proposed framework offers a powerful ability to capture city dynamics and provides valuable calibrations to urban planners in terms of functional zones.
335 citations
Cites background from "Functional regions within the city ..."
...As early as 1970, [29] provided a case study on functional regions within Central London using surveyed data of taxi flows collected in 1962, which is part of the London Traffic Survey....
[...]
TL;DR: In this paper, the authors consider how one of the oldest and most widely applied statistical methods, principal components analysis (PCA), is employed with spatial data, and identify four main methodologies, which are defined as (1) PCA applied to spatial objects, (2) PCAs applied to raster data, (3) atmospheric science PCA, and (4)PCA on flows.
Abstract: This article considers critically how one of the oldest and most widely applied statistical methods, principal components analysis (PCA), is employed with spatial data. We first provide a brief guide to how PCA works: This includes robust and compositional PCA variants, links to factor analysis, latent variable modeling, and multilevel PCA. We then present two different approaches to using PCA with spatial data. First we look at the nonspatial approach, which avoids challenges posed by spatial data by using a standard PCA on attribute space only. Within this approach we identify four main methodologies, which we define as (1) PCA applied to spatial objects, (2) PCA applied to raster data, (3) atmospheric science PCA, and (4) PCA on flows. In the second approach, we look at PCA adapted for effects in geographical space by looking at PCA methods adapted for first-order nonstationary effects (spatial heterogeneity) and second-order stationary effects (spatial autocorrelation). We also describe how PCA can be...
331 citations
Cites methods from "Functional regions within the city ..."
...Other examples of this use of PCA can be found in Goddard (1970) with taxi data, Black (1973) with transportation of commodities, and both Goddard (1973) and Clark (1973) with phone call data....
[...]
...Key historical literature on the use of PCA from a geographer’s perspective includes the work of Berry (1964, 1966, 1968a, 1971), Gould (1967), Hägerstrand (1967), Tinkler (1972), Mather and Openshaw (1974), Goddard and Kirby (1976),...
[...]
TL;DR: The sub-regional structures revealed in this study are more easily interpreted for transportation-related issues than for other structures, such as administrative divisions.
302 citations
Cites methods from "Functional regions within the city ..."
...constrained by linear routes, and thus, the results are unable to fully reflect the urban dynamics. We apply taxi trip data, which have been widely used to investigate city structure since the 1970s (Goddard, 1970), to implement our method. The recent availability of large volumes of taxi Global Positioning System (GPS) trajectories has strongly promoted related studies, including transportation analysis (Fang ...
[...]
157 citations
References
More filters
01 Jan 1968
TL;DR: In this paper, the authors describe a procedure for ordering and grouping cities by the magnitude and direction of the flows of goods, people, and communications between them, based on nodal regions and central place hierarchies.
Abstract: THE PURPOSE OF THIS PAPER is to describe a procedure for ordering and grouping cities by the magnitude and direction of the flows of goods, people, and communications between them. Current theories of nodal regions and central place hierarchies provide the bases for the recognition of region-wide organization of cities into networks. These two theories were developed by students who recognized that the direction and magnitude of flows associated with social processes are indicators of spatial order in the regional structure of urban society. Whether the flow is local and to the city's hinterland, or regional and to the rank ordering of cities, the notion of central or nodal point is dependent upon the levels of strongest associations within the total flow. 1 The present problem is to develop a method capable of quantifying the degree of association between city pairs in a manner that allows identification of the networks of strongest association. These associations may be in terms of .interactions that occur directly between two cities, or indirectly through one or more intermediary cities. The magnitude of the combined direct and indirect associations is measured by an index that is related to certain concepts of graph theory. This index is used to identify the degree of contact between city pairs and it provides a quantitative basis for grouping cities. The resulting subgroups of cities are analogous to nodal regions. When each city in a study region is assigned to a subgroup, it is possible to specify the rank ordering of cities and to evaluate the functional relations of the nodal hierarchy. In this paper, pertinent geograrphic and graph theoretic concepts are discussed and are then used as a basis for deriving the method of isolating nodal regions. Wihile this method is illustrated by the use of intercity telephone calls in Washington state, the techniques are quite general and may be adapted to
351 citations
TL;DR: In this article, Heisenberg's analysis, showing pattern and order where none was seen before, is not the first example of arraying data in matrix form, but it is one of the most famous examples.
Abstract: IN THE history of the development of all scientific fields, a large number of instances can be found where a man has taken information available to all and has reordered it in particular ways to illuminate an area that was previously dark and obscure. Numerous examples spring to mind, ranging from the classical attempts of Linnaeus to order information on plants and animals by imposing a system of classification, to W. Heisenberg's ordering of known facts about atomic structure (B. Hoffman, 1963; Heisenberg, 1930). The latter example is particularly pertinent to the aim of this expository paper, for the ordering of information by Heisenberg was of a particular type. Taking the known frequencies of electrons, which had been previously expressed in the form of a list known as the Balmer-Ritz ladder, Heisenberg rearranged the same information into a square array called a matrix. Operations on the matrix led to some startling and illuminating results. Not only were some theoretical underpinnings provided for the residual half-quantum of energy that remained in an oscillating particle, but some numbers characteristic of the matrices, called eigenvalues, were interpreted as the same excitation frequencies as the Fraunhofer lines, those critical values we can actually see in the spectrum of visible light as thin dark bands. Heisenberg's analysis, showing pattern and order where none was seen before, is not the first example of arraying data in matrix form. Mendelejeff's arrangement of the periodic table in 1869, William Hamilton's discovery in I835 of the power of the quaternion in mechanics, and Laplace's studies in I772 on the perturbations of planetary orbits set a tradition of matrix ordering and manipulation many years before. In a very different sense, and with a very different purpose in mind, ordering the various pieces of an academic discipline in a matrix array has also pointed up pattern and structure that has been useful and thought provoking. J. Berliner's examination of anthropology from the point of view of an economist, E. Hall's
183 citations
100 citations
TL;DR: Computer examples demonstrate the effects of third stage choices on the results of two classifications of geographical data using numerical methods which have been adopted from other scientific disciplines.
Abstract: A number of recent studies have presented classifications of geographical data using numerical methods which have been adopted from other scientific disciplines. Before a classification can begin, however, the investigator must make three sets of decisions regarding the exact process to be employed. Some of his choices are predetermined by the nature of the data and the research design but others may be taken on subjective grounds alone. Computational examples demonstrate the effects of third stage choices on the results of two classifications. The results are analogous to the general geographical problem of regional cores and zones of transition. All methods extract virtually the same core but choice can govern the classification of units within the zone of transition. Classification is thus basically a subjective process, despite the apparent objectivity of the methods employed.
94 citations
Book•
01 Jan 1968
TL;DR: On the occasion of the installation of professors at the University of Lund on 9th December 1967, two inaugural lectures were given under the common title of "The process of urbanization in the perspective of organizational theory".
Abstract: On the occasion of the installation of professors at the University of Lund on 9th December 1967, two inaugural lectures were given under the common title of "The process of urbanization in the perspective of organizational theory." The first of these lectures contained a synoptic presentation of some of the ideas on which are based research projects now in progress on the subject of economic geography at Lund. The present article is a translation of this lecture with the subtitle "Flows of information and the location of economic activities".
50 citations
"Functional regions within the city ..." refers methods in this paper
...Io2 (1965) 26 For a full statement of the methods see N. SPENCE, 'Multi-factor uniform regionalisation of British counties on the basis of employment data for 196I', Reg. Stud. 2 (1968), 87-I04. 27 R. J. JOHNSTON, 'Choice in classification: the subjectivity of objective methods', Ann. Ass. Am. Geogr. 58 (1968), 575-89 28 Euston and King's Cross/St....
[...]
...FUNCTIONAL REGIONS IN CENTRAL LONDON I8I CASETTI, op. cit., and Q. AHMAD, 'Indian cities: characteristics and correlates', Univ. of Chicago, Dept. of Geography Res. Pap. Io2 (1965) 26 For a full statement of the methods see N....
[...]