Bayes’ Rule in determining Oscar Awards
For all the movie lovers out there, the Oscars is one of the events we look forward to every year. We cross our fingers, say our prayers, anxiously sit and wait for our beloved stars to (hopefully) win an award. Thanks to Bayes’ Rule, we no longer have to do that. We can predict who wins best picture using probability! In the article, “And the Oscar goes to…Bayes’ Theorem” by Scott Bronder, Bronder applies Bayes’ Theorem to find the probabilities of winning one of the six main categories. He considers the P(A) event to be the probability of winning best picture, P(bP). He sets the P(B) event to be the probability of the five other categories:
The probability of winning best director, P(bD)
The probability of winning best actor, P(bAr)
The probability of a winning best actress, P(bAs)
The probability of winning best original screenplay, P(bO)
The probability of winning best editing, P(bE)
With those probability of A and B, Bronder utilized Bayes’ Rule: P(A|B)=P(A∩B)/P(B) to find the probabilities for each category. For example, the probability of winning best picture given that the same film won best director is represented by P(bP|bD)=P(bP∩bD)/P(bD). Bronder gathered information regarding the films that were nominated for each category as well as those that won an Oscar using BeautifulSoup and Pandas importing.
The probabilities he ended up with were:
The probability of winning best picture given that film won best director:
P(bP|bD) = P(bP∩bD)/P(bD) = .072/.2 = .36 = 36%
The probability of winning best picture given that film won best actor:
P(bP|bAr) = P(bP∩bAr)/P(Ar) = .0198/.097 = .204 = 20.4%
The probability of winning best picture given that film won best actress:
P(bP|bAs) = P(bP∩bAs)/P(As) = .007/.097 = .072 = 7.2%
The probability of winning best picture given that film won best original screenplay:
P(bP|bO) = P(bP∩bO)/P(bO) = .011/.215 = .0512 = 5.12%
The probability of winning best picture given that film won best editing:
P(bP|bE) = P(bP∩bE)/P(bE) = .034/.197 = .1726 = 17.26%
Source:
https://medium.com/@scbronder12/and-the-oscar-goes-to-bayes-theorem-d5e56d5928a