5 Free (But Not Easy) Ways To Improve Your Data Skills

By Tony Garvan

Working with data is fascinating and rewarding, and the skills to do so are in high demand. If considering a career as a data scientist or engineer, have confidence that the field is wide with opportunity and contains plenty of room for everyone, including those of us not born to rocket science. A lot of courses are out there, and they’re great, but some of what you can do to level up your skills (not to mention attracting attention of potential employers) are things that require independence and creative thinking, as opposed to following the lines of a predetermined course. The tips below don’t require crazy special skills or training (or $$) — what they mostly demand are a stand-alone work ethic and an inquisitive nature.

1. Find a Data Set You’re Interested, Analyze It, and Blog About It.

Doing coursework or kaggle competitions are fine, but a lot of the most important progress in data science happens not in solving the problem, but in defining it. Practice going from the “big” problem to the “little” problem by starting with an expansive question you’re interested in, finding a dataset to answer it, and finding the answers yourself. Interested in crime? Politics? Weather? These days public datasets exist for virtually everything, and trying to get answers out of them is intriguing.

If you’re looking for a place to find interesting data, check out the Kaggle Datasets or Data.gov.

Write up your analytical experience with the data in a blog-style format that will appeal to real people. Try to narrate the analysis into an interesting story: What did you find that surprised you? What was difficult to find out? What new questions were led to? Make the data sing. Or at least make it hum a little tune. If then you post it around social media or relevant subreddits, you’ll grow from what people say. And, besides, it’s fun!

2. Practice Fundamentals Online

Online exercises can be a great way to hone your skills by providing hassle-free setup, well-defined challenges, and instant feedback. HackerRank is a site that offers an online environment to solve a wide array of problems. It’s a great way route to immersion in fundamentals of algorithms, data structures, databases, machine learning, and more. Making progress on some of these concepts with books alone is really hard, but on-the-job experience with such a diverse range of skills can take a lot of time, so facilitate and accelerate the process with by just working through problems online. Don’t worry about getting the hardest ones or ranking the highest: the more hours you put in, the better you’ll be, so don’t frame your efforts in a way that discourages you.

The new site pgexercises.com is a great way to make sure you’re on top of your relational database (specifically postgres) skills. It’s pretty fun to go through as well.

3. Freelance

If you can make the time and stomach the inevitable low-ball pricing, finding work through platforms like Upwork can be a huge investment in your career. Try not to think of it as getting poorly paid for skilled work, but rather as getting a modestly subsidized education. And if you start feeling painfully exploited, remember real work experience ranks higher with employers than education — in interviews I regularly got more interest in a +$1000 side project than in anything I did in my -$70k grad school.

Also, any paid work comes with a free “course” on interpersonal relationships. When you’re solving a real business problem for someone who’s paying you real money, even if it’s $300, the pressures and dynamics are very different than what you would encounter in a self-driven problem set or blog post.

4. Read The Docs

Reading documentation is one of the best ways to get exposure to new technologies — project documentation is often more relevant, practical, and up-to-date than books. Don’t just dig into docs when you’re stuck: if you really want to level up, you can read the docs of popular open source tech cover-to-cover.

Check out the outstanding scikit-learn docs to get a great overview of the current state of machine learning; check out study postgres docs for a great overview of what a modern relational database can do; and delve into MongoDB docs for exposure to NoSQL. Want to get up to speed on distributed processing? The spark docs are pretty slick too.

5. Take a data problem home from work

If you’re employed, odds are your employer works with data in one way or another. The best way you can boost your creative data skills is to find a problem that your business is facing, understand what data your company has or could have, and actually take a stab at finding answers. “Ask for forgiveness and data, not permission,” as the saying goes. This will require considerable initiative and creativity on your part, but the rewards are big: there’s no better way to distinguish yourself in your organization than to demonstrate competence and zeal by taking the initiative to tackling a tough business problem with data.

February 2018