Recently, I read some interesting articles about Uber GPS log data. Some are about a public dataset, which has become invalid unfortunately. But some good guy puts the dataset on Github. You can try it by yourself.
Let us have a look at the articles.
This article shows to us which day in a week people use Uber most and their moving trends. All of the result comes from some Matlab code, which is shared.
A group of 11 judges (including myself) evaluated the teams’ efforts and while most of the teams created some impressive results, they quickly agreed upon the ones they thought were the best. There were five prize categories, and 4 out of the 5 winners used Infochimps data!
That article lists the five lucky teams’ work. You can check them for more insights. I just display some digests here.
MOST ACTIONABLE: Parkalator This is a multi-model parking cost optimization tool for San Francisco residents. It helps drivers decide where to park to save money or whether it’d be cheaper to take a cab.
MOST AESTHETICALLY PLEASING: Marvel Universe Social Graph This is a beautiful display of how Marvel comic characters are related to each other.
BEST DYNAMIC/INTERACTIVE: Disaster Strikes This tool lets you browse and cluster the destruction caused by disasters over the last 60 years. It includes an automatic clustering of nations based on disaster similarity.
PEOPLE’S CHOICE: CuriouSnakes This is a fun tool for dynamically browsing AOL search history. It sometimes reveals unexpected wisdom about human nature.
Uber: Mapping Prostitution and “The Walk of Shame” introduces two results from Uber’s data scientists. Both are interesting and impressive.
PS: Uber has an official blog to list some news about there data. To us it is just a kind of showing off (WE HAVE DATA!!!). You can learn more about Uber there.
After all, I think the company DiDi and Kuaidi should pour their attention on Bigdata. They have some kind of golden resources. If they can provide some services, like guide the free taxis where should go to do a preparation even there is no customer now, I think it should be attractive for the drivers and the companies can make money at the same time. Of course, our researchers can benefit from the public datasets :)