Netflix maps and normalization
January 10, 2010 § Leave a comment
Companies collect detailed information on user’s habits, and sometimes they’re willing to share it with visual journalists. In this case, Neftflix provided The New York Times with movie rankings by Zip code, which the Times turned into noteworthy infographics in print and online on Jan. 10.
A key to making graphics like this work is “normalization.” Let me explain: If you chose to map the NUMBER OF TIMES “Paul Blart: Mall Cop” was rented in each town, that is not normalized. One town might have more people in it than another, so you end up measuring the number of people in each town more than your are measuring the popularity of the movie.
Mapping the RANKING of the movie takes care of this. A heat map showing a movie’s ranking in each location would provide similar results to a heat map showing the number of times a movie was rented in each location divided by the location’s population. (Or at least that’s my theory. I don’t have the data on hand to prove it.)
Another interesting thing to point out about the graphic: The print and online versions were vastly different. I first read the print version in the newspaper’s Sunday Metropolitan section. (Sorry, can’t link to it since it’s a full-page print graphic. I’ll just have to describe.) It showcased a large map of the New York metro region that showed which Zip codes favored “Frost/Nixon,” “Pineapple Express” or “Obsessed.” Below it were six small heat maps of the same region that showed the rankings of six movies.
The online version, which you can see at http://www.nytimes.com/interactive/2010/01/10/nyregion/20100110-netflix-map.html?ref=nyregion lets you scroll through the top 50 Netflix movies nationwide and see heat maps of their rankings in 12 Metro areas.
This is a good example of how visual journalists have to adjust their print and online presentations to fit the strengths of both media.