Congratulations to Aswi: Best Student presentation: AASC2018

BRAG PhD Student Aswi was awarded the Best Student Presentation for her talk at the Australasian Applied Statistics Conference 2018 held in Rotorua, New Zealand on 3-7 December 2018.

Awsi’s abstract for her talk: The impact of covariates on the grouping structure of a Bayesian spatio-temporal localised model is below

Aswi was awarded a certificate and a cash prize. Congratulations Aswi

 

Aswi receiving her award

 

The impact of covariates on the grouping structure of a Bayesian spatio-temporal localised model

Aswi Aswi*, Susanna Cramb, Wenbiao Hu, Gentry White and Kerrie Mengersen
aswi@hdr.qut.edu.au  Queensland University of Technology, Australia

A variety of Bayesian models have been used to describe spatial and temporal patterns of disease, where the data are aggregated at a small area level. A relatively recent approach is the spatio-temporal conditional autoregressive localised model introduced by Lee and Lawson (2016) [1] which allows for spatial autocorrelation between adjacent areas within discontinuous groups. In this paper we use a case study approach to evaluate the impact of covariates on the groups identified in such a model. The study focuses on the influence of climate on annual dengue fever cases in 14 geographic areas of Makassar, Indonesia, during the period 2002-2015, where the climatic factors are measures of temperature, rainfall and humidity. All subsets of the covariates are considered and different spatio-temporal formulations of the model are compared with respect to three metrics: the overall goodness of the fit (Watanabe-Akaike Information Criterion (WAIC)), the group-specific coefficients and the proportion of areas included in the groups. The evaluations are complemented by a range of innovative visualisations of these performance metrics. The results show that inclusion of climatic predictors causes group size and structure to alter in the localised model and that examination of these changes gives greater understanding of their influence. The study also provides more general insight into the behaviour of the Bayesian spatio-temporal conditional autoregressive localised model in the presence of covariates.

Reference
1. Lee D, Lawson A. Quantifying the Spatial Inequality and Temporal Trends in Maternal
Smoking Rates in Glasgow. The Annals of Applied Statistics. 2016;10(3):1427-46.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s