How to move to data science?

[benawabe896]

Hello HN community, I have been working as a LAMP developer for around 8 years now and am thankful to be at the place in my career that I am. (150k at a stable company, low stress, high respect) That being said, I am drawn to data anlysis and data science in general. When I evaluate my experience, I notice that the projects that I enjoyed the most were some sort of analysis. (charts and graphs for internal performance stats, correlation of customer characteristics to product purchases, graphing a customer base to visualize relationship clusters, etc...)

Getting to the point, I'd really like to make the jump from LAMP developer to data analyst/scientist, but I don't really know how. I'm only getting started in performing analysis in a non-brute force way, and have only scratched the surface on things such as random forests, linear regression, and markov chains. While I am an expert in PHP, MySQL, Javascript, etc.., and am very good at working with systems (Fedora, Ubuntu, Apache, VC, CI, etc..), I am a novice in data. Ideally, I'd love to learn on the job, but am open to trading free work for instruction. Will squash bugs for data! :)

Has anyone here made this same type of jump? Would love to hear feedback.

Thanks, Sam

[demian]

There seems two be 2 main fields that you must domain at a fundamental level: Statistics and Business.

Stats, because it is the science of fiding information in data. Get a knack for the basic models, and work your way to the harder and more sophisticated ones.

Business, because you must learn how this information can create value. This facet seems to be often forgotten, but the main point of Data Science is optimizing businesses. Learn about how Big Co. is working, how they tried to solve the problem of Big Data, read about Data Mining, Datawarehousing, ERPs, and other approaches.

There are some other established fileds from where to find great insight, like Operations Research, but these two are the ones you must incorporate in your thinking habits to get the right holistic view. Other fields can even be considered as just extensions of the mindset.

[benawabe896]

Thanks for the reply demian. Did you or someone you know make this type of career switch?

[demian]

Yeah, most of them ex-programmers turned either "Data Scientists" or "Business Analyst/Consultants".

From a career perspective, most of my friends had to prove that their time as programmers didn't "hurt" their potential to think in terms of business. I don't know where you work -you seem to be well compensated so maybie it doesn't apply- but business guys (especially in Big Co.) tend to see programmers as "workers" that "grind" what they are told to. So, besides learning the hard skills, they had to learn how to talk in "business", and to show that they know how and why the company makes money.

Some got moved around in the company they worked as programmers/devs. Others just quitted and got a job somewhere else. It all seems to deppends on your skills, your network, and how well you can sell your potential.

PS: all of them had either engineering or CS degrees.

[pfarrell]

O'Reilly has a deal on a data science starter kit. I don't receive any kind of commission here, but I can tell you, it's quality material.

http://shop.oreilly.com/category/get/data-science-kit.do

[benawabe896]

Very cool. I'm currently going through O'Reilly's "Python for Data analysis", and can speak to it's quality as well.

[scottyallen]

You should put your email in your HN profile so people can email you directly (the email field isn't publicly visible). In the meantime, shoot me an email at [e-mail].

We help businesses better understand mobile apps and app markets through by aggregating tons of data - we have lots of thorny big data problems to solve, and you might be a good fit with your existing skill set for a new project we're about to start.

[one-two]

I am exactly the same way. Look forward to any feedback. Serg

[abhijat]

I have an area of interest which largely coincides with these, I believe, except I am looking to work in ML. I will look at this thread closely myself.

[lazyfunctor]

you may find this blog post by Brad Cross helpful. Unfortunately his blog is not active anymore but it's available on waybackmachine http://web.archive.org/web/20100118082135/http://measuringme...

[abhijat]

Thank you. I have some of the books listed in there but I had a problem with serializing the process of studying them, I tried sort of doing everything in parallel such as studying probability and Tom Mitchel's book together; needless to say that was not going too well.

That link provides a nice sequence in which I should study the books by bringing them all together in an order.

[benawabe896]

I just hope more people see the question.