Last night, I had the extreme privilege of getting to see a group of seventeen budding Data Scientists give the best presentation I've ever seen.
I don't mean the best presentation I've seen from students, or the best presentation I've seen about complex data, or the best presentation I've seen where over a dozen people speak, or the best presentation I've seen created in less than two weeks (although all of those things are also true). This was simply one of the most compelling data stories I've ever seen.
Two weeks ago, me and a colleague walked the Data Science cohort through the Hop Teaming Referral Dataset, which illustrates how patients move through the healthcare system by listing pairs of providers that see the same patients. In short - it's large, messy, and extremely powerful. As a mock consulting project, we asked them to use this data to identify doctors who saw a lot of patients, but didn't share many with the biggest local hospital systems.
Two weeks later, the students came back with results. I've taught enough to know that two weeks is simply not enough time to fully digest novel data, especially when that data is really large and/or complex, and especially when the students have full-time jobs on top of their studies.
But watching this presentation, you could be forgiven for assuming that these students were running a very successful consulting business based around sharing insights from this data. The presentation was that polished and the insights were that profound. When they used complicated models, they presented the data in a way where the results practically explain themselves. When they had to summarize behavior of thousands of doctors, they organized it in an incredibly thoughtful way that would allow any user to quickly understand it. They found very useful and powerful insights such as one local hospital dominating referrals from most doctor specialties, except for Interventional Cardiology. Most single-person presentations don't flow as smoothly that this one that was piloted by one person, shepherded by another, and ultimately involved over a dozen speakers.
These are just a handful of examples out of many stellar options - I urge you to take some time and watch the entire thing. The presentation portion of the video lasts about 40 minutes, and I promise you that you will learn something useful (you can probably skip the last 20-30 minutes unless you want to see me yammer on about how impressed I am - if you're this far into the article I'm sure you already understand!).
I don't mean the best presentation I've seen from students, or the best presentation I've seen about complex data, or the best presentation I've seen where over a dozen people speak, or the best presentation I've seen created in less than two weeks (although all of those things are also true). This was simply one of the most compelling data stories I've ever seen.
Two weeks ago, me and a colleague walked the Data Science cohort through the Hop Teaming Referral Dataset, which illustrates how patients move through the healthcare system by listing pairs of providers that see the same patients. In short - it's large, messy, and extremely powerful. As a mock consulting project, we asked them to use this data to identify doctors who saw a lot of patients, but didn't share many with the biggest local hospital systems.
Two weeks later, the students came back with results. I've taught enough to know that two weeks is simply not enough time to fully digest novel data, especially when that data is really large and/or complex, and especially when the students have full-time jobs on top of their studies.
But watching this presentation, you could be forgiven for assuming that these students were running a very successful consulting business based around sharing insights from this data. The presentation was that polished and the insights were that profound. When they used complicated models, they presented the data in a way where the results practically explain themselves. When they had to summarize behavior of thousands of doctors, they organized it in an incredibly thoughtful way that would allow any user to quickly understand it. They found very useful and powerful insights such as one local hospital dominating referrals from most doctor specialties, except for Interventional Cardiology. Most single-person presentations don't flow as smoothly that this one that was piloted by one person, shepherded by another, and ultimately involved over a dozen speakers.
These are just a handful of examples out of many stellar options - I urge you to take some time and watch the entire thing. The presentation portion of the video lasts about 40 minutes, and I promise you that you will learn something useful (you can probably skip the last 20-30 minutes unless you want to see me yammer on about how impressed I am - if you're this far into the article I'm sure you already understand!).
If you’d like to interact with the data directly, you can find it here. They have even published their code on github so anyone can see how they did it.
Finally, and most importantly, these students are learning so that they can make a positive change in their careers. If you have an opening for a data-focused role in your company, please reach out to the folks at NSS or directly to the students. The Nashville Software School does outstanding work preparing folks for careers in software development, UI/UX design, analytics, and data science. I believe in the quality of their program so fervently that I end up hiring one of their graduates almost every time I have an open position.
If you are as impressed as I am with the quality of their work, please share. Great work needs to be celebrated, and this is a project worth celebrating.
Finally, and most importantly, these students are learning so that they can make a positive change in their careers. If you have an opening for a data-focused role in your company, please reach out to the folks at NSS or directly to the students. The Nashville Software School does outstanding work preparing folks for careers in software development, UI/UX design, analytics, and data science. I believe in the quality of their program so fervently that I end up hiring one of their graduates almost every time I have an open position.
If you are as impressed as I am with the quality of their work, please share. Great work needs to be celebrated, and this is a project worth celebrating.