Mathematical and statistical foundations and challenges of (big) data sciences

摘要

The hype around data sciences in general and big data in particular and the focus either on the potential commercialvalue of data analytics or on promoting its adoption as a new paradigm in conducting research, may crowd outimportant discussions that need to take place about the theoretical foundations of this ‘emerging’ discipline. InSouth Africa, discussions around (or the mere mention of) big data, especially within the National System ofInnovation, often go hand in glove either with the Square Kilometre Array project and astrophysics, or eResearch orcyberinfrastructure. In his excellent essay ‘50 Years of data science’, David Donoho of Stanford University remarks:The now-contemplated field of data science amounts to a superset of the fields of statisticsand machine learning which adds some technology for ‘scaling’ up to ‘big data’. Thischosen superset is motivated by commercial rather than intellectual developments.Choosing this way is likely to miss out on the really important intellectual development ofthe next fifty years. .