Regarding Paul Waterhouse's letter 'Dangers of more data' (The Actuary, October 2014), as of last week, many years after my first stint at university, I am back at University College London studying machine learning.
I think I still have a little while to go before I am sufficiently expert to enter the fray, but one thing that has impressed me is the structured approach to problem solving. While what I have seen so far does not explicitly cover Paul's concerns, it is not too hard to imagine an updated understanding of the Actuarial Control Cycle including both machine learning ideas and the broader issues that Paul mentions. Well, at least, that's what I think after one week. One person who articulates clearly the machine learning approach is Professor Yaser Abu-Mostafa.
The first two lectures in his series ('The Learning Problem' and 'Is Learning Feasible?') are well worth a listen.
And now I have to dash - as my assignment deadline is next Tuesday. Who said that a student's life is easy!