The techniques are built up step by step, from decision trees, random forests, and neural networks, through clustering and dimension reduction, to the tokenisation, embeddings, and transformer architectures behind today's large language models. Beyond the methodological and mathematical background, the seminar concentrates on practical knowledge, suggestions, and assistance for participants' own work: a hands-on block introduces the working environment of a data scientist and follows a complete analysis in Python, from preparing the data to training and evaluating a model. Actuarial use cases from various sectors of the insurance industry illustrate and motivate the procedures and concepts throughout. Equal attention goes to the limits of these methods, such as overfitting, bias, and the black-box problem, and to the judgement that their responsible use requires.
Event informatie
Tijden: 08:45 - 17:00 uur
Locatie: Hotel Casa Amsterdam
Plaats: Amsterdam
Spreker(s): Simon Hatzesberger and Johannes Schupp