The world’s leading publication for data science, AI, and ML professionals.

Five Lessons I’ve Learned As A Meathead To Help Pick Up Data Science

"When you know the way broadly, you see it in all things." – Miyamoto Musashi, The Book of Five Rings

_"When you know the way broadly, you see it in all things." – Miyamoto Musashi, The Book of Five Rings_

Fun fact: those two on the right lost a $20 bet that I couldn't lift this
Fun fact: those two on the right lost a $20 bet that I couldn’t lift this

Over a year ago, I’ve started out my journey into Data Science largely as a means to keep me occupied during the global pandemic in a more productive manner compared to my earlier plan to try to make it as a Twitch streamer (I didn’t have a proper camera set-up). Since then, I’ve completed a few courses, did a few projects along the way, and wrote about my experiences. While others have probably completed much more than that, it’s still some decent progress.

However, as things are starting to resume back to a new kind of normal in my part of the world, it meant the return of some of my other passions, namely lifting really heavy s#*t. While it’s awesome to know I can finally go back to them which I’ve been doing consistently for nearly as long as Lebron James has been playing in the NBA, I can’t help but be reminded of some of the things that I’ve learned in those pursuits being present in this new endeavor.

So, I figured why not let "Uncle Mike" share some of those lessons to help you make progress on this path.

LESSON 1: "Any method will get you there. Just stick to it!"

Look, there are hundreds, if not thousands, of ways and methods that exist in the world to get you where you want to go. Whether it’s diets and training plans to get you to be jacked enough to be The Rock’s body double or lift as much as that giant dude from Game of Thrones, the same can be said with goals relating to Learning data science. For example, getting decent enough with understanding data science to get a job at a FANG company. There are many ways to get up the proverbial mountain.

Original image from Flo Maderebner from Pexel
Original image from Flo Maderebner from Pexel

"You gotta love those ‘free’ timeshare presentations, right?"

Hell, just log onto Instagram, Tik Tok, or YouTube and you’ll find at least 10 people that all did it in various different ways. Sure, there may be some methodology that you’ll probably dig some method more than others, with some evidence showing multi-modal approach being particularly efficacious, but by in large they all work. However, the commonality between each method in relation to its efficacy is rooted in compliance.

Compliance is the science. It plays a big role in a lot of outcomes in things like treatment for health outcomes as well as learning as noted [here](https://files.eric.ed.gov/fulltext/ED504556.pdf) and here. While there is no single way to address this as the factors impacting compliance varies, the key is to be able to recognize all potential factors and provide workaround solutions to make the most of your own situation.

Notice that I didn’t mention anything about efficiency or effectiveness. That’s because what you have available to work with likely isn’t going to be all that efficient or as effective, especially if there’s a lot on your plate. While your rate of progress will probably be a lot slower than others, it’s still progress. So long as improvement is happening, you will still get to your destination. It’s just going to take you a while. Shoot, I know people that are 5 years into their 2-year plan. What’s important is to recognize this to be the case, accept it, and have the resiliency to ride it out until the goal is reached.

LESSON 2: "You’ve got to earn the right to progress."

Let’s face it, somewhere deep down inside of all of us, there is some level of a "Type A" personality in all of us that drives us to want to succeed as quickly as possible. While in that other world it meant trying to get a 315 lbs (144 kg) bench press in a year or two, the data science equivalent would be getting a data analyst/scientist/engineering job ASAP. Or, maybe wanting to get to the point where you can start engineering machine learning models for practical decision making, AI-related automation, or get into the wacky world of deep-fakes with neural networks.

Original image by Ono Kosuki on Pexel
Original image by Ono Kosuki on Pexel

"If you know, you know"

While this mindset is admirable, it can also be a double-sided sword. Notably, this comes in the form of overlooking some of the finer details of things while learning for the sake of progressing as fast as possible. Often this happens after reaching a particular milestone where as soon as this is met, the next one is always in your mind.

It is important to recognize that reaching this achievement equates to capability and not competency. If one mistakes this as the latter, it only serves as a catalyst for stagnation in the learning progress and the eventual abandonment of the venture upon repeated exposures to dilemmas where your current skillset cannot solve for.

Original image by Kajetan Sumila on Unsplash
Original image by Kajetan Sumila on Unsplash

Becoming competent requires one to reasonably produce desired results consistently from its deliberate application under exhaustive experimentation and trials. An example of this would be repeating the data wrangling process through repetition using various data sets (including image/text/numerical data sets) with varying levels of "dirtiness" or introduce a timing element to see for yourself how long it takes for you to complete the process. These scenarios are something that’s commonly done during the hiring process for any data-related position. As such, it would be worthwhile to approach your learning the same way every now and again to see where you are really at.

LESSON 3: "Starting over is always easier when you’ve done it once."

OK, starting anything from zero is always the hardest thing to do. Between not knowing anything and the perceived difficulty to comprehend even the most relatively simple tasks can be both exhausting and demoralizing, it just sucks. So, it’s no surprise that once we get past that "noobie" stage, we never want to experience that again.

Well, sometimes life has a funny way of putting a wrench in that. Whether it’s a thing for work/school, an unforeseen pressing issue that requires your full immediate attention, or being physically out of commission until further notice, we’ll likely lag behind from our original plan of action and our skills regress. Learning data science is no different, it’s just that you don’t get physically injured doing it. Well, I hope not.

Original image by Keira Burton on Pexel
Original image by Keira Burton on Pexel

"This might possibly be the only realistic scenario where you’ll get physically hurt in this pursuit."

The most important thing in these moments is knowing that you can always get back into it. Sure, you might end up realizing that you’ve forgotten more than what you’ve learned and have to do an ungodly amount of Google searches than you care to admit, but that’s alright. There are still things that you can recall that no longer should be a foreign concept to you. You’re not starting back from zero! Consider this the silver lining to an extended layoff of some kind.

LESSON 4: "You can’t make any real progress if you don’t track it!"

Nothing monumental that has stood the test of time has ever been achieved by doing something without a plan and a plan can’t be followed without ensuring certain criteria have been met. Just like those jacked folks that always write in their handy notebooks after each session to track their progress, you need to do the same with your learning journey. While this partly has to do with creating a sense of accountability to sticking to the long-term plan, it also will serve as a log to observe which areas need improvement to allow you to progress to your end goal.

Original image (left and right) by Mikhail Nilov on Pexel
Original image (left and right) by Mikhail Nilov on Pexel

In my case, I record in my log the sort of different data science-related tasks that I accomplished in any given day and record my performance (i.e., ease in the task, time to complete, etc.). After a reasonable period of time, I’ll take a look back at my log and see if there are any less-than-desirable areas for improvement. If so, I can formulate a plan of action that primarily focused on developing that weak area to eventually become a strength.

Given that a machine learning model is only as good as the collected data, the same can be said with your own learning progress.

LESSON 5: "It’s OK to do stupid. Just do it less often."

I believe that a part of going along any venture worth doing, you’ll end up doing pretty idiotic things at some point. Normally this coincides with being in your 20s and greatly overestimating your sense of invincibility, but in the context of learning, it can stem from the deep desire to maximize your learning progress. While it may not seem like it, there are "learning data science" equivalents to trying to get a new bench press personal record while being hungover AF or doing whatever the f#*k this is.

Original image by Mikhail Nilov from Pexel
Original image by Mikhail Nilov from Pexel

The most obvious forms of these are doing consecutive long (6+ hours), deep dive binge learning sessions, or coding complex data projects where you are fuelled by nothing but [insert favorite brand] energy drinks. You know the ones I’m talking about. Where you slowly start morphing into Rami Malek from Mr. Robot.

Now, I’m not going to say don’t do them as there will be times these are needed or that they don’t achieve anything as the desired end goal does sometimes get achieved. Nor will I say that they aren’t satisfying because it definitely can be in the same way that completing a marathon or Iron Man is satisfying. However, I will question, if all of that is really necessary considering the trade-off? Couldn’t we approach all of this to get the same result in the end without going through the Gates of Hades?

Of course, we can. It’s just a matter of being tactful and strategic about it. This means setting up useful habits/behaviors, adequate project planning, and guidelines focusing on productivity and wellness. Some of these include:

  • Set a 90-minute cap to accomplish any given task before moving on with a 15-minute gap in-between tasks
  • Make a rough plan of what can feasibly be done and be conservative with it
  • Spread the cumulative mental stress across a longer period by paying attention to how you feel after each task. If you feel you need an extended break because of burnout, take the mother-f*#king break!
  • Prioritize the quality of work completed over the quantity of work done. More isn’t always better; it sometimes just means more things to fix for your future self.

As a world-renowned spine specialist once said, "There’s only a number of times that you can do these kinds of things before something goes, so pick and choose when to do it and when not to".

There you have it.

My top five simple lessons from my years being a meathead that was being used in this new endeavor. Of course, there are many more. However, I feel these are the ones that hit home the most.

So, what about you, my fellow data science learners? What are some lessons from you that you’ve used in your other passions that you’ve applied to learn this field?

Feel free to comment below and let me know your thoughts.

If you’re interested in connecting, hit me up on my LinkedIn.

Thanks for the read and be on the lookout for some other interesting stuff to come.


Related Articles