These Are the Top 6 Things People Get Wrong When Teaching Themselves Data Science*0qLDudVeAlJhiV8r

Original Source Here

1. They base their progress on other people’s journeys.

When I first started studying data science, I would read every article I could get my hands on about people who had done what I was trying to do: become a self-taught data scientist.

Like many before me and many who will come after me, I devoured these articles that outlined how people without computer science degrees overcame the odds and taught themselves to be data scientists in a mind-boggling small amount of time. The most alluring stories, naturally, were the ones where these people then went on to be hired practically on the spot by FAANG companies.

Articles with sensationalist headlines such as “How I Became a Data Scientist in 3 Months” or “How I Became a Google Data Scientist After Studying Data Science for 1 Year While Working Full-Time” littered my reading list, promising me that if I too “hustled”, that I could achieve my wildest dream of becoming a data science who had achieved career satisfaction and a six-figure salary, all by this time next year.

The glaringly obvious issue here is that my journey into data science was nothing like that of others I had read about. To be fair, only a small percentage of those who enter the data science field will have journeys even remotely similar to those described in those sensationalist articles. In fact, it’s those few articles wallpapered with hope and the glamour of hustle that represent the minority of data scientists who were able to achieve success without a degree or periods of study over several years.

The problem with basing your progress in data science on other people’s journeys is threefold:

  1. Advantages are everything.
  2. Just because it worked for one person doesn’t mean it will work for everyone.
  3. Learning to become a data scientist in anything less than a year ought to be considered a miracle.

First, several unspoken advantages may have helped these phenoms learn data science in a short period of time. The ability to study for eight hours a day, having access to a computer, being able to afford MOOCs and boot camps, or even just being located in a tech hub are just some of the advantages that these people may have had.

Second, what worked for one person may or may not work for you. It’s truly a shot in the dark to attempt to follow the same path to success as someone else. That other person may have a background in tech, they may be able to learn concepts quickly, or they may have contacts that got them their job without even having to interview. All of these cases may not be the same for you.

Third, I applaud anyone who can teach themselves data science in less than a year. Luckily, few out there have accomplished this feat, so I won’t have to applaud for long. The fact is that data science is arguably one of the toughest knowledge fields to dive into and learn on a whim. The unique combination of mathematics, programming, and industry knowledge that you need to absorb to be able to call yourself a data scientist is vast and complex. Therefore, if you can even learn just enough to get your foot in the door, you’re already a phenom.

In short, basing the progress of your journey off of the progress or success of others is a quick way to find yourself discouraged or worse, looking for a quick fix.

2. They learn data science without doing data science.

One of the top complaints of anyone who has ever hired a computer science or software engineering graduate is that they don’t know how to code. While the new hire may have tip-top theoretical knowledge, they’re basically useless once in the office because they can’t write a line of code that has substance or clarity.

The same problem goes for data scientists.

Learning data science is one thing, but applying data science is another. And which is an employer going to care more about?

You guessed it.

A data scientist who can’t analyze data is about as useful as an astronaut who can only land a spacecraft on the moon in theory.

Instead of sitting and listening to your data science lectures passively, you must apply every single lesson that you’ve learned. Memory consolidation is the reinforcement of your comprehension of knowledge that occurs through frequent retrieval of that knowledge. Short term memories of the knowledge that you receive through listening to your data science lectures can only be transferred into your long term memory (otherwise known as the part of your memory that allows you to apply what you’ve learned in the future) through repetition and practice of that knowledge to cement it into your memory.

Therefore, whether it’s trying out the math by hand to see if you can get the same answer, writing a line of code to see if it does what the instructor says it should, or putting all of your knowledge together and trying your very own analysis, the application of knowledge should always be at the forefront of your study.

Here are some resources to get you started:

3. They don’t have direction.

Data science is a massive field that encompasses nearly as many specialties as there are flavors of ice cream in the world.

In the data science community, the debate remains whether or not data scientists should remain generalists or become specialists. With enticing arguments on both sides, it can be difficult to see clearly through the mud-slinging.

However, if teaching yourself data science is the goal, then it may be a good idea to consider specializing in a broad area of data science. What do I mean by this?

Take, for example, me. I am a bachelor of science student attending an online university and essentially teaching myself absolutely everything. I could take a broad degree, studying biology, chemistry, physics, geoscience, the works. However, because I’m teaching myself everything, I wanted to have direction and therefore continuity in my studies. Therefore, I specialized in geoscience.

Therefore, when it comes to learning data science, it’s not a bad idea to specialize in a broad area of data science. As an idea, this could mean focusing more on the data engineer, data analyst, data architect, data scientist, or software engineer, etc. side of things. Each of these areas has their own set of specialty topics yet are broad enough that you won’t be pigeonholed later in your search for jobs.

Here are some resources to get you started:

4. They try to re-think the wheel.

Back in the 1960s, Katherine Johnson and the team at NASA were struggling with developing the math necessary to accurately calculate and execute a change in the trajectory of a space capsule that was required to move from an elliptical orbit to a parabolic orbit in order to bring the astronaut John Glenn safely back to Earth. The insolvable problem was quickly realized when, instead of trying to develop new math to solve the problem, old math was used (Euler’s Method) to be exact.

The moral of this story is that instead of trying to re-think the wheel, Katherine Johnson landed on a solution that already existed. Instead of spending years trying to develop the math that didn’t exist at the time, the space race with the Russians was pushed to new heights by using math that already existed.

In short, aspiring data scientists are always looking for new, phenomenal ways of solving data science problems that are out-of-this-world and unique. The issue with that is that employers are often looking for results in a timely fashion, which means that overcooking the problem unnecessarily when a simple solution is staring you in the face will not serve you well in the job market.

The truth is that you will be admired as a data scientist if you can get clean, actionable results in a reasonable amount of time. This means not trying to find the fanciest or shortest way to solve a problem if a perfectly workable solution is sitting right in front of you.

Here are some resources to get you started:

5. They jump in the deep end without first learning how to swim.

The number of people who jump straight into machine learning, nonlinear statistical models, and deep learning before even being able to code “Hello World” is concerning.

When it comes to data science, everyone wants to immediately get started with the “sexy” side of the job (the data analysis, the model building, the trend predicting, and so forth), but don’t want to deal with the ugly side (learning how statistical models work, understanding programming, making mistakes).

It’s true that to understand data science you need to begin tackling topics that make you uncomfortable, but the fact remains that without a firm grounding in the basics of mathematics, computer science, and industry knowledge, you will fail at being a data scientist long before you ever successfully build a successful machine learning model. In short, everyone thinks they could fly a plane until they’re actually faced with flying a plane.

Instead, if you’re hellbent on learning data science the hard way, one of the better ways to make a quick entrance into the world of data science would be to employ an intense method of study, such as Ultralearning, to first develop your comprehension of the foundations of the subject. That way, you understand the basics of the foundational areas of the field and can then expand your way into machine learning and other more complicated topics.

Here are some resources to get you started:

6. They don’t make a habit of learning.

Learning to learn is an unspoken requirement of the data science journey.

Many over-enthusiastic people armed with a dream of becoming a data scientist who will achieve job satisfaction and a six-figure salary jump straight into their studies without much regard for whether or not they know how to learn. According to these people, who doesn’t know how to learn? Everyone knows how to learn. Yet, in six weeks (or less, depending on how long their enthusiasm can carry them for) these same people will be forgoing their studies for more instantly gratifying experiences.

Not only is learning to learn a requirement of learning data science but making a habit of learning is what will carry you to success.

The trick is to begin by learning how to learn. This means discovering how you learn best and practicing different types of learning to find out what allows you to retain and apply the most information.

The second step is to make a habit out of learning. This means setting aside dedicated study time each day that will be used to learn and practice data science.

Here are some resources to get you started:


Trending AI/ML Article Identified & Digested via Granola by Ramsey Elbasheer; a Machine-Driven RSS Bot

%d bloggers like this: