You want to be a unicorn -- er, data scientist? Three pillars of higher learning promise to make that happen via the magic of the InterWebs
It’s been a while since I covered MOOCs. It sounds vaguely like an epithet, but for those of you who have been hiding in a cave, MOOC means “massive online open course,” which in normal-people talk is “taking courses online.”
It has also been a while since I expressed my derision for the term “data scientist,” but in the last few news cycles these two topics have come together: Three major universities now offer online certifications in data science. What’s interesting is the difference between them.
Three ways to get your data science cert
John Hopkins University is offering a paid-for “specialization” certificate through Coursera. The program seems to center around the R language, specifically learning R programming and plowing through the likes of how to cleanse and load data. It also delves into machine learning and various statistical analysis techniques. Sweat equity, some bandwidth, and $490 earns you this nine-course certificate. If you don’t care about the certificate — you probably shouldn’t — you can take the courses for free.
UC Berkeley is offering a master’s degree in data science with a more traditional application process. It appears to be more the regular “online” offering from a university rather than a MOOC (that is, there aren’t massive numbers of people around the world who can afford the full out-of-state tuition, and it isn’t open, since you have to formally apply). However, as a bonus, one of the courses is co-taught with a doctoral student who looks a whole lot like Niles Crane.
The Berkeley program seems to be a lot more theoretical than the one from Hopkins, but it also hits R and covers MapReduce. More surprising is that it addresses a greater number of current topics, such as Spark. The bottom line is serious coin and you can’t simply clicky-sign-up online — you have to apply to the school. Funny, I always thought they were liberal long-haired types at Berkeley.
For those seeking more traditional technology prestige, there’s MIT’s “certificate of completion” in a course called “Tackling the Challenges of Big Data.” The branding makes me think that John Hopkins is better at marketing, not to mention Web design and writing (MIT subscribes to the “big block o’ text” school of thought). Anyway, the course itself appears to be a comprehensive survey covering case studies, data capture, storage design, MapReduce, Spark, security, parallelism, analytics, visualization — the works! MIT wants $545 for the five-module, 20-hours-of-video course. While MIT is not as liberal with its offering as Hopkins and not nearly as conservative as Berkeley, there appears to be no way to survey it without paying.
Who cares about the cert?
The bottom line is if you don’t have a Java/Linux background, database experience, and maybe basic understanding of statistics, you aren’t going to be supersuccessful in this brave new world. If you aren’t aiming for an academic career, a master’s in data science seems not only premature, but not particularly cost-effective.
Considering that with the background, work ethic, and interest, companies such as mine will pay you well and train you in “data science.” The expense of a master’s degree seems out of line. Moreover, if you have the knowledge, I doubt anyone cares about the certificate. No HR or hiring manager worth their salt trusts certifications — let alone “certificate of completion” — without knowledge verification.
That said, if you’re in a stable job in a company you love and are looking to cross over into newer technology and a more promising field, either the Hopkins or MIT course could be a good way to do it. From a perspective, I can’t help but feel you should know more about storage than the Hopkins course promises, but also using R through a whole project could be a good foundation.
If you have one large burning a hole in your pocket, then grabbing both certs could almost definitely help an existing employer decide to move you over from your day job maintaining legacy code in WSAD to the shiny new data science team. If that doesn’t win them over, you could suggest that your coworkers all complete UPenn’s new course “Wasting time on the Internet.”
Have you taken one of these? Did it help you in some way? If so, I’d love to hear from you.
