|
|
|
|
|
|
by dude_abides
4180 days ago
|
|
|
Some great tips here for budding data scientists: * Form a hypothesis before you start looking at data, else you're susceptible to post-rationalization. * When in doubt, have the raw data available to reconcile with aggregate data. * When publishing results, include your data sources, so that others can verify your findings. |
|
|
This is a really great data analysis. The bottom-line conclusion of "cab drivers who are driving CMT programmed cars are making more money in tips" will definitely cause the Verifone drivers to say "Wait, what?"
The Businessweek article also shows the limitations of the so-called data journalism. The two reporters grabbed the data from the database, made some pretty graphs, got some quotes and called it a day. It took some readers revisiting the data to tease out the real insights.
But that's a bit unfair to the Businessweek guys. They're on a deadline and won't get paid extra for geeking out on the data too much, so their goal is to get the initial news out there to the world. Ben Wellington's interest wouldn't have been piqued if the article hadn't been written in the first place.
I guess what I'm saying is that it's a win all around.