Event Details

This talk is by Philip Jonathan the Chief Statistician for Shell Projects & Technology, and will be speaking about the uses of and challenges in practical applied data science.

Data science and associated fields of machine learning, machine vision, natural language processing and "AI" are seen as "game-changing" technologies in many commercial, scientific, governmental and academic spheres. The ease, size and speed of data acquisition, processing, connectivity and storage, combined with the easy availability of sufficient computational resources, mean that methods for empirical inference can be applied in an ever-increasing range of situations. Whereas the underlying inference problems are often of themselves technically rather mundane, the challenge lies in applying relatively well-known algorithms for statistical learning (which have typically existed in some form for decades) to problems of considerably increased scale, and enabling deployment of "at line" or real time solutions. This talk will outline some of the areas in which the "digitalisation agenda" and data science is impacting an organisation like Shell. Areas where data science is proving most beneficial will be summarised, as will current difficulties. For instance, we will consider the critical challenge of "model assurance" in applied data science, involving the development and implementation of in-situ methods for "algorithm agnostic" assessment and comparison of data-driven solutions to practical industrial problems.