How do I get a job as a data scientist if I have no prior experience as a data scientist? originally appeared on Quora: the place to gain and share knowledge, empowering people to learn from others and better understand the world.
The most common misconception that I’ve come across among data science aspirants and neophytes is that you can learn how to do it just by teaching yourself how to use some specialized tools (like R, scikit-learn, pandas, matplotlib, etc.) It completely ignores the fact that these tools represent just the tiny, visible part of the iceberg, the rest of which is rooted in thorough knowledge of statistics and the scientific method (which is why "Scientist" appears in job titles and there at least used to be a strong preference to hire PhDs).
Putting up code isn’t very useful because it represents a tiny bit of the process required to make something; Kaggle competitions aren’t very useful because they’ve done a lot of the hard work for you already and reward people who know what they’re doing and people who are just willing to try everything for those extra decimal points equally. Very rarely will you find a data science job that tells you exactly what the question is, allows you to build a model on already collected and cleaned data, have it be evaluated by a single numerical measure, and then get to walk away without explaining what you did or what your model implies.
If you want to get experience doing data science, do data science instead of the touristy stuff that Kaggle represents. Identify a question worth asking that can be answered with extant datasets, and devise a strategy to answer it (which is a hell of a lot harder than it sounds!) Once you get to that point, then you can start messing around with tools and building the external-facing parts of your project.
This question originally appeared on Quora. the place to gain and share knowledge, empowering people to learn from others and better understand the world. You can follow Quora on Twitter, Facebook, and Google+. More questions: