A couple of comments here:
For PAXboy;
>>It is cheaper to allow the mistakes to happen and then fix them - than to prevent the mistakes in the first place<<
Well, not really. Take the example of an airliner that is struck by a ground vehicle. The direct cost of repairs may run into several hundred thousand pounds. Then the airline has to lease in another aircraft to run the service. Then there are the litigation costs and the cost in man hours sorting out the problem.
Effect safety monitoring of such hazards actually SAVES money.
For Laurienz;
>>With this in mind is there any mileage in self rostering? Would it be possible in this industry?<<
I think it is a great idea. However, the obstacles are political. Airlines regards their pilots as a resource. They like total control over rostering issues as a way of maximising usage. A system which put the control of the roster in with the pilots will always be rejected by the industry however desirable it may be.
I like the sound of your MSc project. Keep up the good work.