Spatio-Temporal Analytics of Network Data Tao Cheng, James Haworth SpaceTimeLab Team University College London http://www.ucl.ac.uk/spacetimelab GIS Ostrava 2014 - Geoinformatics for Intelligent Transportation j.haworth@ucl.ac.uk; tao.cheng@ucl.ac.uk
http://ww.ukctcs.org/energy/electricalgrids/index.aspx http://www.stardustmoderndesign.com/2012/06/america-revealed.html http://www.amitours.co.uk/blog/amitours-taxi-brainstorm-london-glastonbury/
http://www.ensembl.info/blog/2013/12/16/cancer-researchers-bring-ensembl-to-london/ http://www.theguardian.com/environment/bike-blog/2013/aug/09/cycling-road-safety http://crowdcentric.net/2011/05/can-you-help-us-solve-the-worlds-traffic-congestion/ http://www.chrisspeed.net/?page_id=795
Why Spatio-temporal Analytics? Spatio-temporal data pose unique problems Heterogeneity/Nonstationarity Dynamics Interactions and associations Volume Human Behaviour Different challenge to modelling non-spatial data.
What is Spatio-temporal Analytics? Analytics: Information resulting from the systematic analysis of data or statistics (Oxford Dictionary) Representation Analysis Modelling and Forecasting Spatio-temporal Data Visualisation Clustering Simulation
Space-time modelling and forecasting - The challenge lies in the non-stationary (heterogeneity) and non-linearity of space-time data. Statistical Approaches STARIMA models space-time geostatistical models spatial panel data models space-time GWR How to calibrate the spatiotemporal autocorrelations is the bottleneck. Machine Learning Approaches Artificial Neural networks (ANNs) Self-organizing maps Support vector regression (SVR) Nonparametric Regression The interpretability of machine learning is low Computationally intensive
Challenges in forecasting spatio-temporal data How to: Collect/analyse/use large amounts of data in real time. deal with missing and noisy data on the fly. integrate data from various sources. quantify the effect of network structure on network performance. detect and respond to abnormal (non-recurrent) events. Abnormal congestion event
Space-time clustering - to extract meaningful patterns (clusters) To detect outliers or emerging phenomena (epidemic outbreaks or traffic congestion) Considering the spatial, temporal and thematic attributes seamlessly and simultaneously, and the dynamicity in the data is the most difficult challenge in spatio-temporal clustering Spatio-temporal scan statistics (STSS) sheds lights on this aspect (GAM -> SSS -> STSS) Efforts are needed to improve our understanding of the expected value.
Incident detection and response using Twitter data Tom Wicks
Space-time Simulation - Imitation of the evolution of a process over space and time. Simulation of networks usually based on physical process. Traffic flow theory Bottleneck lies in accurately representing human behaviour. Individual decisions (micro level) lead to large scale effects (macro level). Humans behaviour is not optimal. Agent based modelling is one choice.
Ed Manley Agent-based Simulation Regent s Park Saturation 0 0.2 0.2 0.4 0.4 0.5 0.5 0.6 0.6 0.7 0.7 0.8 0.8 0.9 0.9 1.0 1.0 1.2 1.2 1.5 > 1.5 Hyde Park
Adel Bolbol Fernandez - Understanding Travel Behaviours from GPS Data Logs GPS Data SVM Classification, Spatial Clustering, Map Matching, Spatial Analysis Accuracy achieved > 88% Mode of Transport GPS Testing data: 110 participants in London 2 Months/ participant 20 second collection rate
Space-time visualisation (visual analytics) - explores the patterns hidden in the large data sets using advanced (analytical) visualization and animation static 2D maps 3D wall maps and isosurface (hotspots in space-time) Tools: Visual Analytics and Geovisual Analytics Methods are needed to show evolution and dissipation in space and time simultaneously (e.g. crime or traffic congestion) Still, real-time visualization of dynamic processes is still very challenging due to large volume and high dimensions of the data.
Isosurface for visualising traffic congestion Garavig Tanaksaranond
Wallmaps for visualising traffic congestion Tube Strike- 17:00 6 Sep 2010 21:00 7 Sep 2010 Garavig Tanaksaranond
Garavig Tanaksaranond Space-Time Visualisation of Traffic Congestion Visualising Congestion Build-up in London 3D Wall Map Travel Time Interactive Visualization Tool
Research Frontiers in Spatio-temporal Analytics 1) Modelling and forecasting - nonlinearlity & nonstationarity, abnormal events 2) Tools to capture/illustrate the processes - emergence and tipping points - simulating behaviour (macroscopic properties alter because of accumulated microscopic changes) 3) Spatio-temporal dependence and interactions - impact of activities on the network - interactions between networks BigData empirical theory and testing
Thank you. Questions? j.haworth@ucl.ac.uk standard.cege.ucl.ac.uk