Well, it occurs that the cast of a movie is much easier to obtain than keywords and much less messy. So, I also had scraped the 4 lead actors for each movie of the Beautiful Information awards in order to determine their contribution.

I’ve done it slightly differently than with the keywords. I’ve also taken the budget of the movie into account. So, in order to predict the earnings of a movie, you take the budget, multiply it by 3, remove 8 millions and then add (or remove) the contribution of each star.

Since the movie budget already (mostly ) includes the pay of the lead actors, the way to read this contribution is how much extra these actor should be paid when they appear on film. For instance, each time Emma Stone does her thing, it would be fair to pay her $500k more.

the bang being a modest 500k pay raise.

Likewise, Kirsten Dunst here could be paid an extra 200k per movie, like Rebecca Hall or Rooney Mara. But I mostly felt like posting a picture of Kirsten to stand up against what google autosuggests as keywords when you look her up.

Kirsten, too, should be paid more.

Nicolas Cage, somewhat unsurprisingly,  should refund $2.44m for each movie he stars in.

Actors of movies who’ve done exceedingly well such as Avatar or the Harry Potter or Twilight sagas, movies which, it’s fair to say, owe their success to more than the actors, come out of this over valued. And because of how this is calculated, actors who have costarred with them in less ambitious movies come out undervalued: for instance, Elizabeth Banks or Anton Yelchin who both played with Sam Worthington are “paying” the fact that Man on the Ledge or Terminator: Renaissance haven’t been as successful as Avatar.

Twilight fans take note, though, that Harry Potter actors are considered more valuable.

Now – the data.

Part of it are monthly competitions that run on a curated dataset. Jen Lowe and I are making a team for the current one about the movie industry and we are going to deliver a competitive entry! even our drafts are rocking! But, while we were looking for that one great idea I explored the data.

Part of the dataset is the total box office earnings of over 600 of the movies released in the US in the last 5 years. What I did is cross that list with the user-contributed plot keywords on imdb. Then, I ran a regression on that to find out how much each keyword would generate in the box office? (I only kept those keywords mentioned over 5 times, that’s just over 2500, because else the full list is  20,000+). The full list is at the end of the article.

ok the main motivation for doing this was after seeing the Tree of Life.

Yeah!

Tree of Life was a contemplative family drama, but Terrence Malick turned that into a totally awesome contemplative family drama by throwing in the mix dinosaurs and outer space scenes. I would have liked to prove that a liberal addition of dinosaurs can turn every movie into a better movie, alas, the numbers think otherwise. Movies with dinosaurs will make $500k less in the box office than movies without. For the effort, directors might just as well use dogs instead (movies tagged with “dog” make an extra $4.6m).

and one can see why.

What else can we tell from the numbers?

Believe it or not, nudity will not generate extra profit – movies tagged with nudity will make $2.9m less. Full frontal nudity makes a movie lose even more money. “Partial nudity” or “implied nudity” is ok, though. Now sex is another story. Sex will make a movie get another $6.7m (actors should just keep their clothes on).

Violence is always a hit, too! Explosions generate $6.6m profit, murder is a classic, too – +$2.7m. And “nude woman murdered” is +$1.2m, so if you really need nudity in a movie… Then, you have to choose your weapons – pistol, rifle and rocket launcher make $5.6m, $5.5m and $5.4m respectively. Axes and knives work, too (+$1.4m and +$2.0m) but “knife throwing” or “threatened with a knife” is a no-no (-$2.8m and -$3.5m).

Success in Hollywood starts with a title. Ideally, it should have 2 words, include the name of a character and a place name and punctuation. Then, one character should say the name during the movie. If you pull this out, your movie will make almost $40m extra.

Other very lucrative elements include: a hotel, a gas station, trains, buses, male-female and mother-son relationships, love triangle, and the aforementioned sex and explosions. Also, if you get a character spit on another in the face, that’s an extra $7m.

Caveats:

this is user contributed data so it is quite messy. Some of the keywords are very similar and just because a movie is tagged with one doesn’t mean it’s tagged with the other – like “jew” and “jewish” for instance, “samurai sword” and “katana sword” or many such pairs. And just because a movie is not tagged with a specific keyword doesn’t mean it shouldn’t. Out of the set, there are “only” 135 movies with male-female relationship, compared with 186 tagged with “husband wife relationship”… and regardless, there are too few samples to be really accurate. And now, on with the data!

How to read this: Movies revenues start from a base value, a bit over $43m, and then for each keyword they are tagged with, they can be expected to generate the value in the leftmost column in additional revenue (or lose that much money if that’s a negative value).
Let’s suppose I have a script idea with an alien invasion, giant robots, laser guns, an action heroine and, of course, explosions. Well, I can expect it to make circa $55m.