An ICSC Webinar with George TH Ellison, Professor of Data Science and Director of the Centre for Data Innovation, UCLan.

Predictions occupy a poorly explicated social space within popular and professional discourse, in part because they manifest as theoretical constructs that can be empirically determined (or empirically informed)  or entirely speculative (if not fantastical), or an indeterminate mixture of both. This variability in their potential epistemological origins undermines both the way these are understood (by specialists and lay individuals) and their practical utility/application; not least since the term 'prediction' (and its associated taxonomy of 'risk factors' and 'predictors') can itself further obfuscate their meaning and value.  
   
Given the multiple paths through which empirically- and speculatively-derived heuristics can become real (or 'embodied') through decision-making processes that make these 'self-fulfilling' (or 'self-defeating') prophecies; predictions exhibit perhaps the most extensive 'social lives' of any analytical devices. Getting ahead of the impacts these can have (on our thinking and on our actions) therefore poses an enduring challenge.    

This presentation sets out to explain why the choice of the term 'prediction' to describe statistical tools (such as generalised linear modelling and machine learning) that are capable of accurately specifying past, present and future phenomena extends the 'social lives' of the data on which these are based, in which their strengths (and weaknesses) become conflated with many of the enduring challenges that beset theoretically infused decision-making (and its cognitive recourse to fantastical, latent processes that are not so very different to wizardry, magic, witchcraft or sorcery).    

For more ICSC related events click here