Options arise for expanding data scientists' skills
Data science programs aren't just for universities anymore. Now, data scientists can turn to massive open online courses, online texts and other resources to boost their skill sets.
A number of new programs offer courses, training and opportunities to boost data scientists' skills and certifications. Some of these are modern extensions of existing university programs designed to improve the basic skill sets of data scientists. Many new data science programs, however, provide more ad hoc opportunities to grow data science skills or extend them into other domains.
Data science certifications can increase a data scientist's earning power, as can real-world experience creating different data science applications. The following are some promising programs worth exploring.
Go deep with university programs
University-level data science programs can provide a solid foundation to identify weaknesses in a data scientists' skills base and help them learn complementary tools and techniques. While these programs used to require students to take a few years off, new data science programs are offered online to allow data scientists to earn degrees in their spare time.
"We believe machine learning will be a field in itself for many decades to come -- and if that hypothesis is right, I would shy away from 'become a data scientist in a few weeks' approaches," said Darshan Rawal, founder of Isima, a data science company in stealth mode.
He said he believes programs like the University of California, Berkeley's master of information and data science degree and Carnegie Mellon's Master of Computational Data Science program have the depth and duration to be effective.
Tim Lafferty, director of analytics at Velocity Group Development, said Columbia University's data science certification program also does a good job of covering the discipline as a whole. It includes sections on algorithms, probability/stats, machine learning and visualization. The program is heavier on the development side, which Lafferty said he believes can help data scientists stand out in this rapidly growing industry.
Massive open online courses
A number of massive open online courses have sprung up over the past several years, such as Coursera and edX. These programs, which can help improve data scientists' skills at their own pace, include free offerings, but they charge a nominal fee for certification.
Andrew Ng's machine learning course on Coursera provides an option to earn a certificate from Stanford University, while Kirill Eremenko provides several online courses through SuperDataScience.com.
"Whether just beginning or needing a refresher, Kirill does a great job of taking complex concepts and explaining them in such a way that my 4-year-old daughter would understand," Lafferty said.
Short and focused data science courses
Online courses are great, but some data scientists might prefer a brief but intense dive with live instructors and classmates. For example, Galvanize offers a classroom-based training program to become an entry-level data scientist in seven weeks. The program also pairs students with businesses to work on real business problems.
Unofficial education to boost data scientists' skills
Several new data science programs are springing up on blogs and social media. Kevin Markham created the Data School on YouTube, which provides hundreds of hours of data science content.
Once you are past the point in your life for academic learning, you need access to people in the industry from whom you can continue to learn.
Pat Ryanexecutive vice president, SPR
"Once you are past the point in your life for academic learning, you need access to people in the industry from whom you can continue to learn," said Pat Ryan, executive vice president of the Emerging Technologies practice at SPR, a digital design agency. "After visiting Markham's website, I signed up for the private class, 'Machine Learning with Text in Python.' He recently launched 'Data School Insiders,' which, for a monthly donation, gives you access to ask him and the forum participants questions."
Towards Data Science is a blog with explanations about simple and advanced data science modeling concepts. It also includes links to papers and presentations from top AI and data science practitioners and academics.
"I'm most interested in new tools that democratize data science; provide a graphical, easy-to-use UI; and suggest the best algorithms for the data set rather than going through a multitude of lengthy trials and errors," said Isabelle Nuage, director of product marketing and big data at Talend.
Tools with these programs make data science accessible to more people, like business analysts, to help them benefit from sophisticated and advanced analytics.
Kaggle is another option for exploring what other data scientists are applying to solve particular problems, said Eduardo Franco, data science lead at Descartes Labs, a predictive analytics company. The site includes various data sets and working projects to experiment with, and it provides data science programs to master new skills.
Learning by attacking data science problems
Another approach for building data scientists' skills lies in solving new types of problems.
"I recommend starting with a real data science problem and trying to solve [it]," said Ali Ghodsi, CEO and co-founder of Databricks. "Learning by doing will teach you more than anything else."
Looking outside data science is important, as well.
"I routinely learn from mathematicians, cosmologists, weather modelers and computer vision gurus," Franco said.
These kinds of data science learning opportunities are also being formalized in the Insight Data Science Fellows Program. This new data science program offers a seven-week, post-doctoral program that gives data scientists an opportunity to work on new problems.
"Fellowships help you gain practical experience on projects and skills that [help you] get a foot in the door of your desired industry and build a portfolio," Franco said.
Dive deep with free data science books
Several leading data science researchers and academics have begun posting the full text of their data science books online for free. These books offer data scientists self-paced data science programs in a way that makes it easy to learn about new techniques or to improve basic skills. Examples include:
