The Future of Data Science is Increasingly Entrepreneurial
How I Learned to Become a Data Science Entrepreneur
Let’s face it, one reason we love data science is because it is exploding right now. Every industry from retail to waste management is looking to data scientists to help remain relevant. Translation; it’s a great way to make money.
But if you’re like me you also spent the better part of your professional career learning data science on the side. Through back-alley internet searches for code, blogs, free tutorials, and countless hours of bug fixing on your laptops. Educating ourselves on data science has been tough to come by because the field simply isn’t a good fit for the university model. It moves too fast, there is too much to cover, and few can afford those of us with experience.
In other words, we are self-starters. We have the discipline, determination, and tenacity to go it alone, bang things out till they work, and serve as our own best champions to prospective employers. We’re used to interviews like this:
“Can you organize documents into common themes using NLP? Sure can, here’s a link to my Medium post showing how I did it. What about training a model to detect check boxes in an image? No problem, just shared a Git repo for it.”
At heart, we are entrepreneurs. Entrepreneurs with a reputation for disruption. But breaking into the entrepreneurial world is no easy task. So, in this article, I describe my own experience nurturing entrepreneurship with my data science background. Here we go.
I have always had an entrepreneurial spirit. The idea of creating something that creates value for others, enough value that they are willing to pay you for it, is simply exciting. I was fortunate enough to have a father who was also an entrepreneur. I watched him grow his software company over many long years of dedication, innovation, and attention to his clients’ needs. All this during the rise of tech giants like Microsoft and Apple. A time when the world was just waking up to the promise and value of the entrepreneurial mindset.
As entrepreneurs have become more like rock stars in recent years, especially those who can figure out how to turn a buck, so too has our world become increasingly digital. Arguably the age of entrepreneurs has largely been fueled by the immense digital transformation happening. Greater digitization has made it easier for people to share content. And sharing content has made it easier to disseminate content that educates young entrepreneurs with the skills they need to take advantage of this digital culture.
But with great digitization comes great influxes of data and with great influxes of data “comes great responsibility” (Peter Parker [Spiderman]…I had to). It is our responsibility as entrepreneurs to leverage the availability of data responsibly, ethically, and in ways that help us to build valuable products or services for our ideal clients. To do this we must think about data differently and what better skill set for thinking about data differently than data science?
Data Science Helps Us See Data Differently
As a data scientist with some experience, I love creatively thinking about using those skills for solving problems. In fact, I have spent the better part of the last 15 years of my professional life experimenting, in the entrepreneurial sense, with different applications of data science for solving pretty much any problem I’ve faced.
Here’s one example of what I mean. Not too long ago I was faced with a significant challenge. I found myself in one of those weeks where all my responsibilities in work and life were coming to a head. Like, I really started to get concerned that I wasn’t going to be able to meet some of my most basic deadlines (TPS reports, eating, standing up from my desk…you get the picture).
Adding to the madness was a higher than usual teaching load for a few of the universities I support as contributing faculty. You see, typically I only teach a course here or there with maybe 20 students max. Not this term, however. No, this term I was faced with a few additional courses (“Hey Brandon, can you cover an extra course or two this term? We’ve got some faculty on sabbatical” *sweat beads begin to appear on forehead* *recognizing I’m too nice to say no* *in my most non-committal and cryptic email yet* “Uh, sure.”). But rather than let my anxiety and stress get the best of me, I turned to data science for a solution.
Although the details of my solution are beyond the scope, I will just leave you with this
…I used my years of teaching the same courses and grading the same papers over and over again
…in addition to my meticulous discipline to save all of those graded papers along with my comments in a cloud storage bucket
…as data to build a comment recommendation engine using my data science toolkit.
The engine would consume new papers and recommend comments for each paragraph in the paper based on my past commenting history with confidence scores. Was it a perfect solution? No, but it did reduce my grading time from ~10 minutes / paper to ~2–3 minutes / paper and when you have 40+ students in any given week, that’s a lot of saved time.
But I digress. My point is that we as data scientists have tools that can help us see business problems (or personal ones 😊) differently because we see data differently. The tools we use provide us with a different perspective on how data can be leveraged to solve real business problems. So what are my tips for inspiring entrepreneurship?
1. Thought experiments are just as good as real experiments
The first step to learning how to apply your evolving data science skills to entrepreneurial endeavors is to engage in the ever valuable but often unrecognized step-child to the real thing, the thought experiment. A thought experiment is an exercise whereby a thinker, presumably you, allows the brain to wonder and hypothesize whether a real-world problem might have a data solution. With more and more digitization today, there are more and more business problems that have data to be taken advantage of. The goal is to think through the consequences of data and data science decisions in pursuit of a solution to said problems. Going through this type of exercise helps to inform the development of data science thinking skills that will come more naturally the more you do them. Moreover, being released of any hard coding needs or data requirements you can also explore impossible or even absurd implications that may ultimately lead to greater innovation. Such devices have been around for centuries and their value is well established.
2. But still also experiment
As your thought experimenting skill improves, you will identify real pathways forward for solving problems. Pathways so real in fact that you may even start to envision what your code my look like to get you going. If you find yourself in such a position, don’t forget to engage in some actual experimentation. Despite the value in our thought experiments, the truth is they still only live in our heads. And ideas are very different from reality. Making our ideas reality helps to provide a framework for our ideas, identify potential barriers (technical or otherwise), and motivates our need to continue learning.
Here are a few additional tips about experimenting:
Don’t worry about accuracy or performance.
Don’t be a perfectionist.
Find a platform to share what you’ve done.
3. Get excited…
I have discussed how data scientists often struggle in business because we get so excited about data science as an end in itself
Look ma’ my code worked, model trained, and now I can predict the length of petals on a flower 😊
We fail to recognize that our skills are meant to solve for bigger things. The world doesn’t care about data science, the world cares about keeping track of grocery lists or how long it will take to get to hot yoga in traffic on a cold day. Thus, the more you practice thinking like a data scientist (see point #1) the easier it will be to apply that style of problem solving to something that really excites you. And if it excites you, it probably excites someone else.
4. Find communities, contribute, and network with them
I know you already know this but I’ll say it again, networking matters. I like to think about networking in two spheres; the professional sphere and the client sphere. Focus on both but start with the professional networks. Professional networks may open you up to client networks but you need to get involved with professional networks beyond just data science.
Find communities of both data scientists and other entrepreneurs. Doing so helps expose you and your skill set to other professionals who may have access to client networks you didn’t before.
For example, I knew I was good at applying data science tools to processing documents with some intelligence, I even wrote a tutorial on the subject, but what I didn’t know was who needed a solution like that.
I mean, I knew lots of companies might be interested (after all, what company doesn’t deal with documents?) but I didn’t have any contacts at those companies who would help me to develop the solution appropriately.
My solution, network with other entrepreneurs who faced this or similar problems and who have access to a client world that I was unaware of.
My point is, communities are powerful (I started one in part for this very reason *ruthless self-promotion notwithstanding*) and client communities (audiences, if you will) are essential to successful businesses. Thus, learning how to interact and network with those communities can significantly propel you well along the way to success.
5. Did I say start small? Let me re-emphasize
Take advantage of how you learned data science. Most of us first learn data science on our laptops. We download Python or R in some flavor, we traverse examples of code to do things using data that fit nicely within the confines of our operating systems.
Undoubtedly, we have all dealt with the memory errors, slow training times, and dependency conflicts that arise as we punish the CPU processors of the tiny machines sitting in our laps. Frustrating as it may be, being able to successfully relieve memory with down sampling or distributed processing, improve training times with shallower networks or larger convolutional filters, and managing dependency conflicts with virtual environments or containers makes us ever more suited to develop proofs of concept (POC’s) at a fraction of the cost to help future clients buy in to our vision.
The ability to solve small is useful, valuable, and attractive to potential business partners because you know how to keep costs low while providing exposure to skills that otherwise cost a lot of money to leverage.
So what are you waiting for? Trade your skills for equity and start flexing your entrepreneurial muscles.
Like engaging to learn more about data science? Join me.