How I Became a Data Scientist Despite Having Been a Math Major

Read as: how I became a data scientist after taking an appropriate and helpful major

The problem with not having a math background is that you will succumb to all the people who throw around buzzwords but don't know what the hell they're talking about. You need to be able to recognize (and assert authority over) someone whose first thought is that "support vector regression and a bayesian net" is the right way to go. Or, to put it nicer, refine their thought into something useful.

The problem with not having a software and "grungy coding skills" background is that you aren't able to efficiently reproduce/check/verify claims being made which are often wrong or misleading. A recent example I had of this is someone studying how an algorithm behaved on a subset of interest of a particular population, and they didn't even bother to check that the subset of their data was statistically large enough to draw any conclusions (only 17 records out of thousands). Needless to say this example also failed on the reproducibility and reuse fronts.

The problem with both in academia (even at polytechnic non research schools) is that they don't know enough about each other.

Question for the data scientists out there: is it more straightforward to become a data scientist from a software engineering background or from a math background? Data science is a varied field, but from what I've heard a lot of the work is munging data (which is a fairly easy task for a skilled software person).

Also, I'm glad the author added the twitter endorsement to the article --I've been considering creating a twitter account, and the fact that the author found it so useful helps with my decision to devote time to creating an account.

The perfect data scientist -- imho is great at massaging data and has a ton of real life experience with real data; So somebody on the interface of statistics and computer science is best.

Computer scientists mostly don't know about sampling and make easy problems hard. Statisticians mostly fail at simply getting the data in a format they need.

Perhaps PhD math programs don't hold your hand through teaching data science, but it isn't a horrible way of preparing you for a (first) job in data. Color me impressed when a history major picks up data science!

Begin with Andrew Ng's mooc at coursera, to get the basics and the feel of it.

I've completed the very first his course in December 2011.