Lead Supervisor: Francisco Rowe, University of Liverpool
Other Supervisors: Danushka Bollegala (Computer Science) ; Dani Arribas-Bel (Geography)
Contact Email: firstname.lastname@example.org
Partners: University of Liverpool
Start Date: October 2017
There are many applications for spatio-temporal data analysis for solving real-world problems. In the context of a Big Data system, spatio-temporal data are often utilized along with other types of data. In Big Data world, generally there are two computational analytical tasks in processing and analyzing spatio-temporal data (data preparation and cleaning is not considered as a computational task in this proposal). The first task is spatial-temporal data engineering and exploratory data analysis, that is about using various methods, operators and predicates to transform spatial data into a shape suitable for further analysis. In addition, the spatio-temporal analysis might generate a new set of spatio-temporal data product. It may include descriptive and exploratory data analysis. The data engineering and exploratory tasks are most of the time important initial steps in further complex spatio-temporal data analysis. In addition, descriptive and exploratory data analysis often are all that can be done in complex datasets. The second task is about using geostatistics, statistical learning, machine learning, simulation, etc. to model a process, event, activity, behavior, etc. which often results in describing or predicting a process, activity, event, behavior. While having major importance in many types of Analytics, most spatio-temporal data analysis suffer from lack of scalability. The Lack of scalability prohibit utilizing spatio-temporal analysis along with other new types of data in a big data ecosystem. This is why there is no major open source and reliable big data technology for spatio-temporal analysis. In addition, some applications of spatio-temporal data need real-time or near real-time response. Therefore existing spatio-temporal data analysis need to be integrated into existing big data platform (via importing and modifying the source code, redesign of algorithms, etc.). More importantly, the scalable spatio-temporal analytical framework needs to be tested and validated with a large set of real-world data. An objective of this proposal is to create a scalable framework for spatio-temporal analysis. The framework also can act as a guide for making methods of spatial analysis scalable. Another objective of this proposal is that the developed framework must be tested and validated for a real-world application. This can be a specific application like disaster management, or resource targeting.
Reference number LV03
Deadline for applications – 30th April 2017