Semester 2

Date 
Presenter 
Title 
Abstract 
10 Sep 
Assoc Prof Peter Howley
The University of Newcastle

Control charts using Bayesian methods 
[Click to show or hide] 


[YiFan's supervisors are Dr. Peter Howley and Dr. Frank Tuyl]
Peter's interests have seen him involved in several areas of research the past few years, these have included control charts using Bayesian methods, assessing the valueadd of tertiary institutions, the use of clinical indicators in the hospital accreditation process, testing the effectiveness of educational programs, considering relationships between body parts and sex or height, as well as the development of a national poster competition for secondary school students. Peter will give an overview of these activities, which are at varying levels of development.
YiFan has just completed his first year as a PhD student with Drs Peter Howley and Frank Tuyl as supervisors. YiFan’s work focuses on the creation of control charts using Bayesian methods for the monitoring of clinical indicators. YiFan will outline the new chart being tested, the current and intended comparisons with existing charts and his future work.

11 Sep 
Distinguished Professor Noel Cressie
University of Wollongong

Multivariate geostatistics 
[Click to show or hide] 


Multivariate geostatistics is based on modelling all covariances between all possible combinations of two or more variables at any locations in a continuously indexed domain. Multivariate spatial covariance models need to be built with care, since any covariance matrix that is derived from such a model has to be nonnegativedeﬁnite. In this talk, a conditional approach for multivariate, spatialstatistical model construction is given. Starting with bivariate spatial models, its connection to multivariate models deﬁned by spatial networks is given. A bivariate model is fitted to a minimummaximum temperature dataset in the state of Colorado, USA. This is joint research with Andrew Zammit Mangion, NIASRA, University of Wollongong.

17 Sep 
Mr Muhammad Khan
The University of Newcastle

Transmuted Weibull distribution with covariates regression modelling to analyse survival data 
[Click to show or hide] 


Survival experiments are conducted in engineering and biomedical applications to evaluate the effect of a treatment on the distributions of lifetimes of people or things. In this research we introduce and study the transmuted Weibull distribution with applications. This new class of distribution is obtained by using the quadratic rank transmuted map (QRTM) technique. We also propose logtransmuted Weibull distribution and logtransmuted Weibull regression model for survival data. In this presentation we consider the comprehensive treatment of the mathematical properties of the TW distribution including expressions for the quantile, moments, moment generating function, incomplete moment, probability weighted moment, geometric mean, harmonic mean, Beta, Shannon and Rényi entropies, mean deviation, Bonferroni and Lorenz curves, mean residual life and the rth moments of order statistics. This research incorporates regression modelling to analyse survival data for the transmuted Weibull distribution. We also propose a log TW regression model of a location and scale form, derive the moments and discuss maximum likelihood estimation.

30 Oct 
Garth Tarr
The University of Newcastle

Interactive and data adaptive model selection with mplot 
[Click to show or hide] 


This talk introduces the approach to model selection based on the concept of model stability (Meinshausen and Bühlmann, 2010; Müller and Welsh 2010). We present the mplot R package which provides a collection of functions designed to help users visualise the stability of the variable selection process. A browser based graphical user interface is provided to facilitate interaction with the results.
We have developed routines for modified versions of the simplified adaptive fence procedure (Jiang et al., 2009) and other graphical tools such as variable inclusion plots and model selection plots (Müller and Welsh, 2010; Murray et al., 2013). We also propose extensions to higher dimensional models using via bootstrapping lasso estimates and incorporate robustness to outliers via an initial screening process (Filzmoser et al., 2008).
While the focus to date has been on linear and generalised linear models, we are currently working on expanding the set of methods available to encompass mixed models and robust regression models. We will give an overview of what has been achieved and discuss areas for future research.
REFERENCES:
Filzmoser P, Maronna RA and Werner M (2008). Outlier Identification in High Dimensions. Computational Statistics & Data Analysis 52(3), 1694–1711. DOI: 10.1016/j.csda.2007.05.018.
Jiang J, Nguyen T & Rao JS (2009). A simplified adaptive fence procedure. Statistics & Probability Letters, 79(5), 625629. DOI: 10.1016/j.spl.2008.10.014
Meinshausen N, and Bühlmann P (2010). Stability Selection. Journal of the Royal Statistical Society: Series B (Statistical Methodology) 72(4), 417–73. DOI: 10.1111/j.14679868.2010.00740.x.
Müller S & Welsh AH (2010). On Model Selection Curves. International Statistical Review, 78, 240256. DOI: 10.1111/j.17515823.2010.00108.x
Murray K, Heritier S and Müller S (2013). Graphical tools for model selection in generalized linear models. Statistics in Medicine, 32, 44384451. DOI: 10.1002/sim.5855
Tarr G, Müller S and Welsh AH. mplot: Graphical model stability and model selection procedures. Preprint.
