17 October 2019

Research pick: Your next favourite movie - "On the application of Bayesian credibility theory in movie rankings"

Can statistical and probability mathematics invented in the eighteenth century help fans choose their next favourite movie? A new study published in the International Journal Operational Research suggests that it might be so.

Palash Ranjan Das of the Department of Applied Mathematics at the University of Calcutta, in West Bengal, and Gopal Govindasamy of the Madras School of Economics, in Chennai, Tamil Nadu, India, explain how they have coupled movie choice with Bayesian credibility theory. Credibility theory is a branch of actuarial science devoted to quantify how unique a particular outcome will be when compared to an outcome deemed as typical. Thomas Bayes for whom the Bayes Theorem is named was an English statistician and philosopher who formulated a new approach to understanding chance and probability in the middle of the eighteenth century long before the arrival moving pictures and many decades before the notion of computer software that might assess the chances of a given movie suggestion being one a viewer might enjoy.

Bayesian credibility theory was initially developed to assess risk. However, the team in this current work has used it to rate and rank movies available from an online movie database based on user votes.

Das, P.R. and Govindasamy, G. (2019) ‘On the application of Bayesian credibility theory in movie rankings’, Int. J. Operational Research, Vol. 36, No. 2, pp.254–269.

No comments: