Dataquest’s Active Curriculum: How We Teach Data Science
There are lots of ways you can learn data science, both online and off. But at Dataquest, we take a unique approach that’s based on our own experiences and on years of data from hundreds of thousands of students completing millions of screens.
In this article, we’re going to explain what our Active Curriculum is, and why we teach the way we do.
Our Active Curriculum in action.
What is the Active Curriculum?
In a nutshell, there are three things that set our Active Curriculum apart from other kinds of data science education:
- Carefully-planned content
- Hands-on learning loop
- Data-driven optimization
We’ll examine each of these three things in more detail to explain why we do them, and why we think this approach reflects the most effective way of teaching data science skills.
One of the things that separates our Active Curriculum from others is that it’s carefully planned from the ground-up by in-house experts.
Content that’s based on real-world data work.
Our content team is a mix of former teachers and former data professionals (including a few PhDs). But when deciding what course content to create, we start by reaching out into the broader community.
We analyze hundreds of job postings from all over the world, so that we can teach what’s most in-demand. We also speak with working data scientists and other experts in the field to get more information about the skills that students will really need to work with data in a professional context.
This focus on teaching real-world skills is also baked into the courses themselves. As you learn new programming and statistics concepts, you’ll be challenged to solve realistic data analysis and data science problems similar to those you’ll encounter in your workplace.
One major frustration in online learning can be the gaps between courses that force you to chase down knowledge elsewhere.
Course sequences with no gaps.
Because our courses are almost all written by internal authors, we also consider how each course fits into our existing course catalog as we write it.
Although you’re free to take courses in any order you choose, each of our learning paths is a carefully-designed sequence of courses that allow you to start as an absolute beginner with no prior experience writing code.
When you finish each course, you’ll be able to jump right into the next one without having to waste time chasing down prerequisites or Googling to fill in gaps in your knowledge. Because we control and plan every aspect of the curriculum, we don’t have any of these gaps.
This careful course sequencing also allows us to use later courses to help you review and practice the skills you picked up in courses earlier in the sequence. This allows us to take advantage of the spacing effect, helping ensure you’re retaining what you learn as you move through our learning paths.
Hands-On Learning Loop
One of the strongest principles underlying our Active Curriculum is that to learn effectively, you need to regularly apply what you learn. This isn’t just a theory, there’s quite a bit of science to back it up. A 2014 meta-study, for example, found that students in STEM courses who weren’t regularly applying what they learned were 1.5 times as likely to fail.
The learn > apply > feedback cycle.
Our interactive platform is designed to get you hands-on with what you’re learning as quickly as possible. On each screen, you’ll read about a new concept and then immediately be challenged to apply it by writing code that’s then checked by our answer-checker. Studying at a typical cadence, it is unlikely that a student would go even five minutes on our platform without writing code.
When students compare Dataquest to other similar learning platforms, one thing we hear a lot is that Dataquest ‘actually makes you think’ (here are twoexamples of folks who recently told us that). That’s by design. Each step of the way, we want to challenge you to do what you’ll have to do in a professional setting: write real code to solve a problem.
We believe that having a really quick feedback loop — you learn a little something, you quickly write code to apply it, and the answer-checker tells you whether you’ve gotten it right — makes learning more engaging and more effective. It also makes learning at your own pace and fitting study sessions into a busy schedule more practical.
But we also know that real-world data science work doesn’t happen in tiny chunks. That’s why each of our courses ends with a guided project. These projects simulate realistic data science work. They challenge you to synthesize and apply everything you’ve learned up to that point, while still offering some guidance when you need it to ensure you don’t get frustrated. And they help you build up a portfolio of data science projects you can draw from when it’s time to apply for your next job.
Why no videos?
The lack of videos is one of the biggest differences between our Active Curriculum and most others. We could probably write a book on why we don’t teach data science with videos, but here’s a quick rundown of some of the biggest reasons:
- Any student will see their level of focus change over the course of a study session. It’s natural that our brains can’t maintain a completely consistent focus and comprehension level for an extended period of time. If you’re learning by reading, these little differences don’t matter. You’ll naturally adjust your reading pace to account for these shifts. But videos play at a uniform speed, meaning that for many students, there will be times when the video feels too slow, and others where it feels too fast. This can result in boredom and/or missed content.
- Nobody gets everything right on the first try. When you have to go back and review something, searching through video can be frustrating and time-consuming compared to a simple Ctrl + F.
- Watching people code can be tricky. It’s easy to see someone else do something and feel like you understand it, only to realize later when you try to apply it yourself that you didn’t grasp it as well as you thought.
- Videos can be tough on non-native speakers (even with subtitles) and on students with internet bandwidth restrictions or slow connections.
- Videos, unless they’re very short, slow down the learn > apply feedback loop.
Additionally, although these aren’t problems with videos themselves, a lot of video learning platforms lack code-running capabilities, and thus test students with multiple-choice and fill-in-the-blank quizzes. We don’t think this is an effective way of teaching — if you’re not actually writing code, you’re not really applying what you’re learning.
The best way to learn to write code is writing code, not watching someone else write code.
Real data, real problems, real code.
Because our focus is on teaching job-ready skills, all of our courses use real-world data sets and challenge you to solve realistic data science problems by writing real code. It might be cliche, but we believe in the old coaching adage: you play like you practice.
In other words, if your goal is to work with real data sets to solve real data science problems, your learning and practice should be using real data sets to solve real data science problems. That’s why even Dataquest students who’ve never written code before will be working with a real-world dataset within minutes of starting their first course.
Expert Data-Driven Optimization
Another thing that makes our Active Curriculum unique is that it isn’t static. We’re always adding new courses to our paths, but we’re also tracking a lot of data and constantly soliciting feedback so that we can keep improving the content we already have.
Using data to teach more effectively.
Our platform collects a lot of data about how each of our courses, missions, and screens performs. This allows our in-house course authors to quickly spot when there’s a problem with one of their courses.
For example, if an unusually high number of code runs on a particular screen are failing, that might indicate the screen is too advanced for its position in the course, and needs to be modified or moved.
Course authors frequently dig into this data to make course optimizations that might include changing exercises, tweaking explanations, and shifting content around. The end result is a course curriculum that’s highly optimized, and constantly improving.
At Dataquest we love data — no surprise there — but we also regularly solicit qualitative feedback from students about our courses. Student feedback can help us identify areas for improvement that might not be obvious from the data, and it can be another valuable tool that helps our course authors optimize their courses for student success.
This constant optimization is possible because almost all of our courses are written in-house, by full-time Dataquest employees.
This ensures that someone is responsible for each course. If a problem comes up, the course author is available immediately to solve it. If there’s a way a particular course can be improved, the course author will spot it as part of their regular review of how courses are performing.
It also means that we can be sure all of our course authors are keeping up with the latest in educational research and in the data science industry. We have regular team discussions about new pedagogic research, and we also track industry trends so that we can implement new skills, libraries, packages, workflows, and processes into our courses when they become popular in the data science world.
Regular new course releases.
Of course, in addition to optimizing our existing courses, we’re also constantly working on new courses to add more value to our existing paths and to create new ones. In 2019, for example, we launched a dozen new courses and two totally revamped paths (our Data Analyst in R and Data Engineering paths).
We don’t plan to slow down anytime soon!
In short, there are three things that make Dataquest’s curriculum special:
- Our carefully-planned content
- Our quick hands-on learning loop
- Data-driven optimization to make good courses great