Student research opportunities
Spatio-temporal Big Data Analysis and Computation with Applications in Agri-environmental Informatics
Project Code: CECS_1028
This project is available at the following levels:
Honours, PhD
Keywords:
Missing data, low quality, mobility, non-Gaussian processes, misalignment, Big Data
Supervisor:
Dr Warren JinOutline:
Sensors, especially wireless sensor networks, are broadly applied for, e.g., environmental/earth monitoring, area monitoring and industrial monitoring due to the ease of deployment and their minimal impact on the environment. There are several challenges for predicting the space-time surface of the environmental variables that are being only monitored by sensors at points. These challenges include a large number of sensors; ever larger sensor locations caused by sensor mobility or redeployment; heterogeneity of the sensor distribution; incomplete and missing data. To address these issues under the hierarchical Bayesian modelling framework, this project aims to further develop an efficient and effective spatio-temporal model, ARFSA and/or spTDyn. The former incorporates lag one auto-regression for temporal dependency and a large-scale plus local-scale spatial dependency structure for spatial dependency. The latter is able to handle the spatial and/or temporal misalignment of response variable and explanatory variables.
Goals of this project
- Develop, implement (in C and R) and test a suite of methods for large scale real-world data: directions may to use diagonal matrix in the place of sparse matrix for fast computation, dynamic data structure to handle mobile sensors, more general temporal structure, temporal forecast
- Possibly extended an R package
- Document source code, and possibly distributed as an R package to the public.
Requirements/Prerequisites
- Familiarity with C/C++ and the script language, ideally, R computing language
- Basics of related knowledge, like time series, or spatio-temporal modelling, Gaussian processes
- Interest in solving real-world problems
Student Gain
A student working in this project can expect
- to learn state-of-art of spatio-temporal modelling techniques
- to be involved in developing cutting-edge techniques to handle real-world environmental challenges while working with a research group delivering great science and innovative solutions for Australian society and economy;
- Software package development experience
- Stronger R and C++ programming skills, that will be
valuable for future statistical data analysis or data mining - State-of-art of time series analysis, clustering, or spatio-temporal techniques with applications;
- Real world problem solving;
Background Literature
- R free software for Statistical Computing
- spTimer: Spatio-Temporal Bayesian Modelling Using R
- Gaussian Processes for Machine Learning. Carl Edward Rasmussen and Christopher K. I. Williams
MIT Press, 2006. ISBN-10 0-262-18253-X. - Efficient and effective Spatio-temporal analysis of data collected by large sensor networks. Jin et al 2015. Under refinement
- Khandoker Shuvo Bakar, Philip Kokic and Huidong Jin (2015). “A spatio-dynamic model for assessing frost risk in south-eastern Australia.” Journal of the Royal Statistical Society: Series C (Applied Statistics). In press
- Khandoker Shuvo Bakar, Philip Kokic and Huidong Jin. “Hierarchical spatially varying coefficient and temporal dynamic process models using spTDyn.” Journal of Statistical Computation and Simulation. In press.
- More information could found at his ANU home page or CSIRO staff page (open to CSIRO people only)
CSIRO (www.csiro.au), as Australia’s national science agency is one of the largest and most diverse research agencies in the world. It operates large multi-disciplinary research teams. By doing a project with CSIRO you will have access to world class facilities and be able to work alongside CSIRO scientists while you are enjoying generous personal development and learning opportunities.
