Vizualizing 5000 IMDb Movies

Getting started Movie Madness Face the Film Actors and Directors About

Do the number of faces in a movie poster matter?
Source: Kaggle: 5000 IMDB Movies

What is this?

One of the most interesting portions of this data set is that the scrapper also had a facial recognition algorithm. The algorithm would inspect each movie poster and count the number of human faces in the poster. Above is a visualization trying to explore this face count.

Profile We start off by comparing the number of faces in a movie’s poster to the movies earnings. However, the y-axis can be changed, by clicking an icon on the left, to compare IMDb score and Facebook likes again as well. The color and bubble size still represent a movie’s earnings and you can still hover over a bubble for more details on the film or click the bubble to be taken directly to the movie’s IMDb page.

From a first look, one can see that as the number of faces in a movie poster increase the earnings drastically decrease! This could just be because most movies only have a few faces in a poster (if any), but it could show that theaters should avoid putting faces on posters of movies. When we look at the casts’ Facebook likes we also see that as the number of faces goes up the number of likes goes down quickly. However, when we look at the IMDb score we can see that as the number of faces increases the score seems to be converging around 7.5. So if you want a movie that makes little money, but has a decent IMDb score then consider adding more faces to the movie poster.



Previous Next!



University of San Francisco: CS 360, 2017