Ομιλία Τμήματος Στατιστικής και Ασφαλιστικής Επιστήμης, Πανεπιστήμιο Πειραιώς 17/3/2023: “Too much of a good thing: Model Selection and the Fukushima Nuclear Plant Disaster”

ΠΑΝΕΠΙΣΤΗΜΙΟ ΠΕΙΡΑΙΩΣ
ΣΧΟΛΗ ΧΡΗΜΑΤΟΟΙΚΟΝΟΜΙΚΗΣ ΚΑΙ ΣΤΑΤΙΣΤΙΚΗΣ
ΤΜΗΜΑ ΣΤΑΤΙΣΤΙΚΗΣ ΚΑΙ ΑΣΦΑΛΙΣΤΙΚΗΣ ΕΠΙΣΤΗΜΗΣ

ΠΡΟΣΚΛΗΣΗ

Σας προσκαλούμε στη διαδικτυακή ομιλία του Constantin T. Yiannoutsos, Professor of Biostatistics in the Department of Biostatistics, R.M. Fairbanks School of Public Health, Indiana University, USA, η οποία θα διεξαχθεί την Παρασκευή 17 Μαρτίου 2023, ώρα 16:00 μέσω της εφαρμογής MsTeams, με θέμα:

Too much of a good thing: Model Selection and the Fukushima Nuclear Plant Disaster

Abstract: “All models are wrong, but some are useful”. According to GEP Box, the exercise of model selection involves capturing the most salient features of the data. A useful model will permit, precise prediction of future events (out of sample prediction). By contrast overly complicated models will capture mostly the variability in the data; What is called the bias-variance dilemma. We present an example here of a situation where over-interpretation of the data may have led to a catastrophe of historic proportions, the Fukushima nuclear plant accident. We reanalyze data from earthquakes in the area and try to determine whether the parameters that went into the construction of the power plant were correct or were affected by an over interpretation of the available data.

Για να συνδεθείτε στη διαδικτυακή ομιλία επιλέγετε
https://teams.microsoft.com/l/meetup-join/19%3ac2b84132af2140dabd5355ea486de1a4%40thread.tacv2/1678779252193?context=%7b%22Tid%22%3a%22d9c8dee3-558b-483d-b502-d31fa0cb24de%22%2c%22Oid%22%3a%22845270c1-6fa1-493b-b3af-3ebabd95a79e%22%7d