Big Data Analysis and Data Mining

Biography

Biography: Fionn Murtagh

Abstract

Geometric data analysis allows for “letting the data speak” and integrates qualitative and quantitative analytics. Scope and potential are major in many fields. Case studies here are large scale social media analytics, related to an area of social practice and an area of health and well-being. The interesting survey of Keiding and Louis, “Perils and potentials of self-selected entry to epidemiological studies and surveys” points to very interesting issues in big data analytics. My contribution is in the discussion part of this paper. Through the geometry and topology of data and information, with inclusion of context, of chronology and of frame-models, we are addressing such issues of sampling and representativity. The case studies to be discussed in this presentation are related to mental health and to social entertainment events and contexts in the latter case with many millions of Twitter tweets, using many languages. Particular consideration is given to use and implementation of our analytical perspectives. This includes determining the information content of our data clouds, and of mapping onto Euclidean-distance endowed semantic factor spaces, as well as the ultrametric or hierarchical topology, that is characteristic of all forms of complex systems.