How We Hacked Binomial Distribution ...

... in a bar in Hong Kong.

December 22, 2014

It’s another sizzling day in Wan Chai.

My technology partner for 10 years, Pierre Boquet, and I recently moved from France to Hong Kong to drive the technology efforts for our company FIRST FINANCE Institute, an enabler of online-based certificate programs for prestigious business schools worldwide, including Tsinghua SEM in Beijing and HEC Paris.

Still a bit jetlagged and seeking as much air conditioning as possible, we settle on a bar called The Typhoon to relax and think about an online learning data visualizaiton challenge that we have decided to take on. The challenge reads:

We invite creative coders everywhere to a learning technology challenge. Your mission is to create an interactive visualization of the binomial distribution suitable for students who are learning the topic of probability.”

As we have been tasked with the daunting challenge of converting our own company’s 17 years of training in financial products into a digital cloud-based format, we are more than ready…

In fact, developing interactivity for online courses, fostering social interactivity, and learning by doing is our obsession---and we’re always looking for ways to brush up on our skills.

Data visualization, in particular, is an asset we think that all online course developers ought to use. The right, engaging, and visually interactive element can lead students to explore more, and one hopes, learn better.

Based upon how we approached and completed the challenge, we present five lessons for creating interactive visualizations.

1. Start with the most basic questions, but then be progressive
What are you trying to visualize? Why? Assume, in many cases, that a learner might know very little (if anything) about a particular concept. I myself had to relearn this important lesson.

With just over a month left to submit our entry for the competition, and too many ideas to consider, I emphatically suggested to Pierre that since all binomial distributions look pretty much the same we should have two rather than one. My rationale was that students should be able to see how binomials compare when playing with input parameters.

If you were to open two books about statistics and look up a chapter on binomials and compare graphs, you would find that they are very similar, at least in shape. While the input parameters can be very different, what matters is the scale. Yet, this only becomes obvious when a learner can play around, something only an interactive tool can properly provide.

If you do not understand the above, you are not alone. Pierre took me one step back by making me consider the most basic questions a learner might have: What is a probability? What does success (for example, the chance of a winning hand) mean?

We then decided that we needed to illustrate basic components, the very definition of the binomial, before considering the broader concept of probability.

That is a lesson I would emphasize to all developers who are considering creating interactive visualizations: Start your story at the beginning. Specifically, give a learner the basics before moving into something more elaborate. A tool might be engaging, beautiful, and well designed, but it will have little value if the learner has no idea why they are using it.

2. Find your creative pairing
As illustrated above, when developing an interactive visualization, it is beneficial to work in a team with members who bring different perspectives.

With a strong mathematical background, I had jumped to a conclusion about comparing laws and running numerous scenarios. On the other hand, as a self-taught coder, Pierre questioned my academic approach and stressed the necessity to illustrate the basic components first.

Our entry was not the result of his or my brains, but of a dialogue and creative energy that Pierre and I have established through our many years working together.

3. Sometimes technology really is the answer
We believe that the practice of blended learning (combining the best aspects of face-face training with the best aspects of online education) can provide an optimal learning environment.

Online course developers should always keep that in mind. Use technology wherever it really adds value to explaining a learning concept---or when it encourages exploration.

In our particular example, data visualization allowed students to play with various scenarios and to see for themselves the law of great numbers at work. While it might be obvious, especially if you are technology driven, only consider technology-driven solutions when they are the best---if not only---way to teach a concept.

4. There is not one good and definite answer to a problem
One advantage of taking part in the challenge has been to see the ways that others came up with solutions to the same problem. One solution had a particularly good approach to Gaussian; another used a coin analogy rather than the more typical throwing dice trope.

Moreover, having multiple approaches is not just good for developers and faculty, but also offers students more choices in how they might best explore a complicated topic.

In the MOOC- and open access- era, it is often all about sharing, comparing, and experiencing. Faculty and students alike can enrich a course through constant feedback (what works, what doesn’t work), and sometimes even by offering their own contributions.

A MOOC is open in more ways then one: it is open to students, open to developers, and open to challenges.

5. The power of open source education is just now being realized
As members of the edX open source MOOC community, we have already seen the power of collective work (with ourselves concentrating on ORA questions and analytics).

Consider this learning challenge as an ideal example of how to improve learning through smart crowdsourcing. A problem is posed. Submissions are sent in by technology teams around the world who are working with the same infrastructure. And ultimately, a group of French Ex-Pats living in Hong Kong wins the top prize.

The real prize for us is that we expect that many of the finance faculty we work with will seek to incorporate both our own solution to binomial distribution and other various solutions and tools in their upcoming classes.

Most important, thousands of students taking online courses will be taught something in a completely new and innovative way.

(Note: EdX just announced the Dashboard Challenge, a data-driven competition to design a Student Analytics Dashboard. The competition is hosted by Databits – an online community of coders, engineers, and data scientists who develop creative solutions through data visualization.)

Fabrice Demichel is General Manager chez FBM.


Back to Top