Dr Michael Fop is a statistician currently working as a Lecturer/Assistant Professor in the School of Maths and Statistics in UCD, and as of 2020 is the programme director of BSC and BA in Statistics. In conversation with him I learned what his life as a statistician is like and some of the projects that he is working on.
Statistics has the convenient nature of being very cross-compatible with other fields of study in science and humanities. Dr. Fop and others in his profession have the ability to work from anywhere with an internet connection, which is useful during the current climate of a pandemic. Dr. Fop did his bachelor’s degree in Business, Economics, and Administration in Perugia, Italy. During his time in Italy, he developed a keen interest in statistics. This was fostered with the help of a lecturer in the university who encouraged him to successfully apply himself to a master’s degree in finance and statistics in Perugia. Equipped with this knowledge and experience, he decided to travel to Ireland to pursue a PhD in Statistics under Professor Brendan Murphy, from which he continued into a Post-Doctoral position before starting his lectureship in 2018.
Like all statisticians, Dr. Fop has his favourite models. For him, these are models that are used to find groups or classifying objects known as mixture models, and admits that “Not all models can work for applied problems”. However, the creation of sub-models and other avenues of work can then be applied to deal with a particular problem. His research focuses on the “development of statistical models and methods for the measurement of complex and higher-dimensional data”, characterised by its considerable magnitude that can be analysed in "statistical models used to select set variables which are more informative for clustering and classification purposes”
The research he undertook during his PhD was with physiotherapists in St Vincent’s University Hospital, who were specialists able to provide the data and give explanatory background, to help classify different types of low back pain. In this research, there were three different types of low back pain that could be classified under a general scheme, which gave rise to variables that allowed data analysis to extract patterns from the data. Developing the model, implementing, and coding the statistical model are all the steps required in data analysis interpretation, making sure the results make sense, and finding bugs in the code.
A current project involves work on a social networking platform that has the base definition of being able to extract information from your followers to create a linked network. The work of a statistician is to identify and analyse these patterns, which means that most research they are involved in can be in many overlapping fields. Fop referenced a famous quote from a statistician that resonates with him; “As a statistician, you get to play in everyone’s back yard”. The implications of this paints the possibility of a diverse life’s work with the ability to work on projects that come about through a mixture of interest and chance.
A project that came about through both of these, is his personal work on analysing data on members of the Northern Top terrorist network which was active in South Asia and Malaysia in 2000-2010. Using data released about these members he can map “If they participate in the same bomb attack or the same training session or if they went to school together or same religious place”. This allows him to “summarise relationships that were happening at that time.. [and find].. Potential groups of people that interact together”. He is working in collaboration with colleagues on this network data after seeing a topical news report, and subsequently found the data online which he used to model the data to make inferences.
Due to his interest in the field of astronomy and astrophysics, Dr. Fop has also been working on a model in conjunction with UCD School of Physics to group gamma-ray bursts and characterise them by certain attributes. Gamma-ray bursts are extremely energetic explosions that have been observed in distant galaxies and research on them is used to advance the current knowledge of events in these distant galaxies. The measurements of these bursts have some uncertainty attached to them which is akin to an error bar commonly added to measurements in science. By modelling this statistically, Dr. Fop is able to improve the accuracy of their data by incorporating this error.
Statistics is found hand-in-hand with programming and computer science, with programming there is a “set of sequential steps and you have a problem you have to solve with multiple ways to solve it and you find the most efficient and accurate way to solve”. To implement the analysis of his data Dr. Fop develops his own software as; “all the data analysis models that I develop need to transform into a software”, with his language of choice being the statistical programming language R. His most popular software is the R package Mclust which is a package for clustering and classification, that in the last 6 months has been downloaded over 2 million times. This shows both the popularity of the package and how clustering applications are increasingly being utilised for data analysis, and the paper describing its use and the technology behind it is within the top 100 most cited papers in the last decade in the field of mathematics. Fop believes that “If I developed a method and I believe that this method works and it is useful [then] I need to create a software which implements the method and can be used by other researchers and a bigger community to provide a bigger impact to my research”
Fop is currently teaching multiple modules in UCD that revolve around Machine learning and Artificial Intelligence (AI). Machine learning merges statistics and computer science and is fundamental to the bubble of hype that comes with AI. He does not believe that the hype around AI is substantiated currently as we don’t have the computing complexity to make the AI of sci-fi movies as; “a machine can only do what you teach to it" and currently “we don’t have an answer to what is AI”. Additionally, there is a bias in AI as it is trained on the data given to it which is not always representative of the entire data collection. This means that it is not real intelligence but rather an advanced form of teaching-and-application.
Dr. Fop’s academic research was not affected too much by COVID-19 but he did have to change his approach to work at home, whilst adjusting to the new reality. After UCD ‘s abrupt change from a blended learning approach, it has led to more uncertainty with teaching, and teaching from home is not the same as it misses the in-person interaction. But as he admits “I just need a pen and paper to write down equations and my laptop to code 4.25” and this allows for a lot of flexibility and stability in his work.
Dr. Michael Fop is a statistician currently working as a Lecturer/Assistant Professor in the School of Maths and Statistics in UCD, and as of 2020 is the programme director of BSC and BA in Statistics.