Department: Mathematics and Statistics
Room: Fylde B53
Email: Email This Person
I am a first year PhD student at STOR-i, Lancaster University. My research is in the area of statistical modelling, specifically looking at models for low-count time-series. I am supervised by Dr Nikos Kourentzes (Lancaster Center For Forecasting) and Dr Peter Neal (Department of Mathematics and Statistics, Lancaster). Prior to this, I received a BSc in Mathematics from the University of Cardiff, and an MRes from STOR-i.
So what is my PhD about? In a nutshell my aim is to improve the existing state of knowledge on modelling and forecasting low-count time-series. The integer, non-negative and low-valued nature of low-count time-series present interesting but complex modelling challenges. Standard models for normal, continuous time-series, such as the popular ARIMA models, violate the properties of low-count time-series, and hence different models are needed.
My research looks at integer models to address these issues. My primary aim is to adapt and innovate these integer models (e.g. Poisson regression, INARMA, HMM) to comply with the properties of low-count time-series. For these ‘new’ models, secondary aims include the development of a classification scheme, and also a comparative study of forecasting performance against standard ‘ad-hoc’ methods e.g. exponential smoothing. Other potential avenues of research include multivariate extensions and accuracy measures for low-count data.
Please see my personal website for more information.