Ten Things Online Training in Data Science & Analytics Fails To Deliver
And Why It Matters More Than Anything You Will Actually Learn
Many people use online training to prepare for a career in data science and analytics. Or at least, they try…
Online training, on the surface at least, holds tremendous promise of democratizing and empowering the masses to learn data science and analytics. And yet, a full decade after its arrival, demand continues to go completely unmet. Each year more and more aspiring analysts enroll in online programs and yet few actually land DS&A opportunities. But this article is not about online training’s failure to deliver on a job or opportunity, it is about something much deeper.
When it comes to online training in analytics, two languages always lead the charge — R & Python. In twenty plus years of running global analytics divisions, I never once hired a single analyst for either of those two skills.
I did once try to hire someone with SAS skills (a competitor / precursor to R), but ended up giving the role (as SAS administrator) to an internal transfer. If visual basic is the precursor to Python, you can add two more exceptions…
Twenty year, like 200–300 hires, and only a small handful of times it mattered. And, even then, only to add an admin or automation specialist to a team of analysts. Online courses simply emphasize the wrong things. Or put differently, online courses emphasize the things easiest to teach via online courses. So what should they emphasize?
1) SQL
83% of job openings in DS&A require SQL and yet it is often not listed in Online Course descriptions. This is doubly true when the title is Data Science. SQL is a requirement for success in this field. Don’t avoid it. Learn it!
2) Data Discovery & Organization
You can’t do data discovery or organization of any real value using sample data sets and open sources. Real-world data is dirty, imperfect, and down right messy. You need to learn how to navigate it, organize it, and develop it.
3) Client Discovery
Analyst typically need to define the problem first. It is not simply handed to them. Some organizations can, but even then not all the time. Analysts need to know how to conduct a client interview, to read a P&L, and perform business discovery. None of these are easily delivered in video format at scale… good luck finding them.
4) Requirements Gathering
Analysts perform requirements gathering activities that are unique to their field. Analytic processes are not data management, engineering, development, or what falls to most project managers. Even if one of those other roles is responsible, analysts must learn how to advocate and educate.
5) Adversarial Relationships
Rounding out the more interactive skill sets that are hard fits for online courses, are adversarial relationships. Analysts nearly always encounter these. Learning how to defend your techniques and recommendations or to counter poor ones espoused by others is critical to success. Does Khan Academy have one of those?
6) Question Forming / Hypothesis Development
If you complete your online course and still think it is appropriate to ask who, what, where, when, and why questions — you need to get a refund. More than anything else, analysts define and develop questions. These questions drive hypotheses and form models. Nothing is more critical.
7) Experimental Design
Analysis is a science, but the scientific method does not always apply. Business world conditions are not the same as laboratories. This makes experience in experimental design and testing a must have. Even university courses struggle in this area, and those are priced to have scale…
8) Cognitive Bias
One of my favorite questions for any management candidate — “Please name one of the cognitive biases and describe its impact for me.” I have been known to end interviews three blinks and a silent pause after that question. If you don’t understand cognitive bias — don’t call yourself an analyst or a scientist!
9) Translation & Buzzword Bingo
Analysts have an interesting relationship with confidence. It is both a necessity and a pathway for confirmation bias (a good answer for #8). One of the more common things to trip up young analysts and erode their confidence is the abundance of different names, brands, and labels used in various fields of analytics. Ex. Data Science is just Statistical Analysis with more automation…
10) Communication, Support, and Persuasion
If you want to be a successful analyst, you need to be able to teach. Perhaps more than any other skill, this is so difficult to do well in online course offerings. There are ways to do it — but it is a rare course set-up that can do it well. I have seen a few MBA programs accomplish this. Of course, they had a $100K price tag!
They ten skills make up far more of an analysts day than R or Python. Far more than machine learning. They run far deeper than statistics or capstones. In fact, in the image above only Data Processing, Transformation, and Visualization would complete (and they shoved them all in one segment!). So is there any hope for an online or virtual training system that doesn’t brake the bank but does prepare you for a successful career in DS&A?
We think so. Our platform uses simulation and AI to make that happen:
Good luck. And thanks for reading!
