This month’s event was a showcase of talks presented by three young statisticians from the Vic Branch.
The first presentation was by PhD student Jiadong Mao on methods for analysing streaming data. Streaming data are collected sequentially over a potentially infinite time period requiring real-time data estimation. Jiadong is developing nonparametric estimation approaches for streaming data that are computationally fast (easy to update) and adaptive to nonstationarity. He demonstrated an application of kernel density estimation to satellite data.
Next up was PhD student Rushani Wijesuriya who performed a comprehensive simulation study to compare different multiple imputation methods for handling missing data in three-level data structures (e.g. Naplan data where there are repeated waves of data collection from students clustered within schools). Rushani found that approaches which impute the missing data using a multilevel model performed better than simpler methods that impute data at a single level using a wide format.
Our final presentation was by PhD student Ravindi Nanayakkara who introduced us to cosmic microwave background data, which capture leftover radiation from the Big Bang. The data are complex requiring models to represent random fields on a sphere. Ravindi evaluated model fits using real and simulated data.