Age ratios

For all movies where the age of actors and actresses are available, we compute the ratio of the average of actress ages and average of actors ages. Therefore, a ratio above 1 means that on average, actresses are older than actors. Looking at the distribution of these age ratios for the movies passing the Bechdel test and not passing, there seems to be a higher mean when the movie passes the test. The difference in distributions is also confirmed by a Kolmogorov-Smirnov test. We can conclude that a movie is more likely to pass the test if actresses are older than actors. However, note that for the majority of movies, the ratio is below 1, which means that actors are older most of the time, no matter the outcome of the test !

Genre Network representation

Movie genres were very scattered, and very detailed. Doing Machine Learning on more than 300 different genres would lead to overfitting, and this feature had to be reduced before being used. To do so, it was assumed that certain genres appeared very frequently together, and that it was possible to group genres into clusters. A network representation was made, linking two films when they appeared together in a movie.

move the mouse over the network to see how the genres are linked together

The network is very dense, and there are no groups of film genres that can be distinguished into communities. Genres do not have the same level of detail, and some very general genres, such as romance or action, are at the center of the network and are linked to many different subtypes, which are on the periphery of the network. This explains why the clustering was discarded to reduce movie genre dimensionality.