James Haworth

Inferring hybrid transportation modes from sparse GPS data using a moving window SVM classification

Understanding travel behaviour and travel demand is of constant importance to transportation communities and agencies in every country. Nowadays, attempts have been made to automatically infer transportation modes from positional data, such as the data collected by using GPS devices so that the cost in time and budget of conventional travel diary survey could be significantly reduced. Some limitations, however, exist in the literature, in aspects of data collection (sample size selected, duration of study, granularity of data), selection of variables (or combination of variables), and method of inference (the number of transportation modes to be used in the learning). This paper therefore, attempts to fully understand these aspects in the process of inference. We aim to solve a classification problem of GPS data into different transportation modes (car, walk, cycle, underground, train and bus). We first study the variables that could contribute positively to this classification, and statistically quantify their discriminatory power. We then introduce a novel approach to carry out this inference using a framework based on Support Vector Machines (SVMs) classification. The framework was tested using coarse-grained GPS data, which has been avoided in previous studies, achieving a promising accuracy of 88% with a Kappa statistic reflecting almost perfect agreement.

Spatio-temporal autocorrelation of road network data

Modelling autocorrelation structure among space–time observations is crucial in space–time modelling and forecasting. The aim of this research is to examine the spatio-temporal autocorrelation structure of road networks in order to determine likely requirements for building a suitable space–time forecasting model. Exploratory space–time autocorrelation analysis is carried out using journey time data collected on London’s road network. Through the use of both global and local autocorrelation measures, the autocorrelation structure of the road network is found to be dynamic and heterogeneous in both space and time. It reveals that a global measure of autocorrelation is not sufficient to explain the network structure. Dynamic and local structures must be accounted for space–time modelling and forecasting. This has broad implications for space–time modelling and network complexity.

Non-parametric regression for space-time forecasting under missing data

As more and more real time spatio-temporal datasets become available at increasing spatial and temporal resolutions, the provision of high quality, predictive information about spatio-temporal processes becomes an increasingly feasible goal. However, many sensor networks that collect spatio-temporal information are prone to failure, resulting in missing data. To complicate matters, the missing data is often not missing at random, and is characterised by long periods where no data is observed. The performance of traditional univariate forecasting methods such as ARIMA models decreases with the length of the missing data period because they do not have access to local temporal information. However, if spatio-temporal autocorrelation is present in a space–time series then spatio-temporal approaches have the potential to offer better forecasts. In this paper, a non-parametric spatio-temporal kernel regression model is developed to forecast the future unit journey time values of road links in central London, UK, under the assumption of sensor malfunction. Only the current traffic patterns of the upstream and downstream neighbouring links are used to inform the forecasts. The model performance is compared with another form of non-parametric regression, K-nearest neighbours, which is also effective in forecasting under missing data. The methods show promising forecasting performance, particularly in periods of high congestion.

Improved Methods for Fire Risk Assessment in Low-Income and Informal Settlements

Fires cause over 300,000 deaths annually worldwide and leave millions more with permanent injuries: some 95% of these deaths are in low- and middle-income countries. Burn injury risk is strongly associated with low-income and informal (or slum) settlements, which are growing rapidly in an urbanising world. Fire policy and mitigation strategies in poorer countries are constrained by inadequate data on incidence, impacts, and causes, which is mainly due to a lack of capacity and resources for data collection, analysis, and modelling. As a first step towards overcoming such challenges, this project reviewed the literature on the subject to assess the potential of a range of methods and tools for identifying, assessing, and addressing fire risk in low-income and informal settlements; the process was supported by an expert workshop at University College London in May 2016. We suggest that community-based risk and vulnerability assessment methods, which are widely used in disaster risk reduction, could be adapted to urban fire risk assessment, and could be enhanced by advances in crowdsourcing and citizen science for geospatial data creation and collection. To assist urban planners, emergency managers, and community organisations who are working in resource-constrained settings to identify and assess relevant fire risk factors, we also suggest an improved analytical framework based on the Haddon Matrix.

Mapping Spatiotemporal Patterns of Disabled People: The Case of the St. Jude’s Storm Emergency

Emergency management can greatly benefit from an understanding of the spatiotemporal distribution of individual population groups because it optimizes the allocation of resources and personnel needed in case of an emergency caused by a disaster. In practice, however, vulnerable population groups, such as people with disability, tend to be overlooked by emergency officials. Tasks such as identifying people who are vulnerable in an emergency generally are approached statically using census data, without taking into account the spatiotemporal dynamics of disabled people’s concentrations as observed in large metropolitan areas such as London, United Kingdom. Transport data gathered by automatic fare collection methods combined with accessibility covariates have the potential of being a good source for describing the distribution of this concentration. As a case study, data from the peak of the St. Jude’s Day storm in London on October 28, 2013, were used to model the within-day fluctuation of disabled people, employing discrete spatiotemporal variation methods. Specifically, Poisson spatiotemporal generalized linear models were built within a hierarchical framework, ranging from simple to more complex ones, taking into account spatiotemporal interactions that emerge between space-time units. The performance of the resulting models in terms of their ability to explain the effects of the covariates, as well as predict future disabled peoples counts, were compared relative to each other using the deviance information criterion and posterior predictive check criterion. Analysis of results revealed a distinct spatiotemporal pattern of disabled transport users that potentially could be used by emergency planners to inform their decisions.