Join us for three short talks by current PhD students Jiadong Mao, Rushani Wijesuriya and Ravindi Nanayakkara.
6:00pm – Talks – Theatre 1
, Basement of 207-221 Bouverie St, University of Melbourne
7:15pm – Drinks – Prince Alfred Rooftop & Bar, Carlton
(SSA Vic members receive a free drink!)
Talk 1: Jiadong Mao – Nonparametric estimation for streaming data
In the era of big data, huge volumes of data are continuously collected in time-varying environment (termed streaming data). A model for streaming data should be fast to compute and adaptive to the changing nature of the data. These two objectives are hard to achieve simultaneously. Conventional statistical methods often assume that all data have already been collected and stored in the computer memory. Existing models for streaming data, are mostly proposed by computer scientists and only address the computational challenge. Without knowing their theoretical properties, it is hard to predict when these methods will succeed.
We develop computationally efficient models with theoretical guarantee, with more flexible nonparametric models, and investigate their theoretical properties using infill asymptotics. Computational properties of the models are shown using computer simulations and then applied to some real data examples to show their power for modelling real-world problems.
Talk 2: Rushani Wijesuriya – Multiple imputation in three-level data structures
Multilevel data with three levels of hierarchy are common in health research studies. A common problem in such studies is the presence of missing data and often handled with multiple Imputation (MI). To our knowledge there are only two MI implementations that are specialized for imputing missing data in a three-level setting (one within R and the other in the stand-alone software Blimp) and a lack of sufficient guidance for practitioners regarding the settings for which each of these approaches is appropriate. We investigate the performance of alternative MI approaches for handling three-level incomplete data by means of a simulation study under a number of different scenarios. Based on a case study from the Childhood to Adolescence Transition Study (CATS), we compared a range of currently available multilevel MI methods designed for single-level and two-level data combined with ad hoc approaches, such as the use of dummy indicators (DI) for school clusters or a just another variable (JAV) approach to repeated measures, in terms of bias and precision.
Talk 3: Ravindi Nanayakkara – Stochastic modelling and statistical analysis of cosmic microwave background data
The Cosmic Microwave Background (CMB) is the radiation from the universe since 380,000 years from its birth. In 2009, the European Space Agency launched the mission Planck to study the CMB thoroughly. The aim of the mission was to verify the standard model of cosmology using a detailed resolution of observations and to find out fluctuations from the specified standard model of cosmology. The main statistical model used to describe the CMB data is isotropic Gaussian fields. A random field can be defined as a stochastic process; indexed by a spatial variable. The Rényi function plays a central role in multifractal analysis, since the multifractal formalism in the theory of random cascades can be understood in the sense of the Legendre transform of the Rényi function. For the Rényi function on the sphere, there are three models where the Rényi function is known explicitly. They are Log-Normal model, Log-Gamma model and Log-Negative-Inverted-Gamma model. Our research aims to check the Gaussianity of the CMB Radiation data collected from the Planck mission. We discuss the statistical properties of random fields on spheres using high frequency asymptotics for angular spectrum.