The Fastest Way to Learn Data Science

The Fastest Way to Learn Data ScienceIt's not about what you know it is about what you can doRebecca VickeryBlockedUnblockFollowFollowingApr 30Photo by Tim Mossholder on UnsplashWhen I first started writing blogs about data science on medium I wrote a series of posts describing a complete roadmap for learning data science.

I am largely self-taught in data science and over the last few years have, through trial and error, found some excellent ways to learn in a fast, efficient way.

I am a stickler for efficiency and will when trying to achieve any goal aim to find the fastest route there wherever possible.

In the following post, I am going to share my top tips for accelerating your learning in data science or indeed many other subjects.

Have goals you care deeply aboutTo keep momentum in learning you need goals you care very deeply about.

I am extremely passionate about data science and my goal was to become a data scientist but that wasn’t my main driving force.

I want to impact the world around me, in a positive way, with data.

So my main driving force was to be able to do impactful things with data science.

This resonates so deeply with me and was the single most important thing driving me forward in my learning.

You need to find something that you are really passionate about and set your goals around that.

Keep them in mind always, refer back to them often, this will help you to keep razor sharp focus and move forward very quickly.

It’s about what you can doThis leads onto my next point, and I really feel strongly about this, it is not about what you know it is about what you can do.

Data science is a very broad subject you will never know everything.

It is very easy to get lost learning the theory behind every model or all of the maths you might use up front.

The key though is to focus on learning what you need to be able to do practical things with data science.

So it is fine to dive straight into building a machine learning model with a widely used library.

You can learn the theory later.

The chances are that once you have built something and got it to work your natural curiosity will lead you to understand more of the theory behind it.

Ideas are worth nothing unless executed, Derek SiversThis particularly applies if you want to become a data scientist working in a business.

In business, value vs effort is of paramount importance, and so being able to find the most efficient possible route to building and deploying a model is extremely important.

In business, it doesn’t matter so much that you know everything about advanced calculus, for example.

What is important is that you can build a good model in a reasonable time frame, and of course know enough maths to evaluate, understand and improve performance.

Roadmap monthly reviewWhen I wanted to accelerate my learning in data science I created a roadmap of skills I needed to master.

Alongside this list, I gave myself a rating, along the lines of, never used, beginner, intermediate, advanced.

At the beginning of each month, I would set myself a few skills to focus on.

What this meant was that whenever I had a little spare time I didn’t have to think about what to do.

I just needed to pick something from this list and start working.

I knew instantly what to focus on.

At the end of each month, I would score myself against each skill.

This gave me an idea of what I needed to develop next.

But also served as a helpful reminder of how I had far I had progressed each month.

You don’t have to complete MOOC’sI learn best by doing and found that video-based courses did not work particularly well for me.

The vast majority of these large courses are video lecture based.

I dabbled with MOOC’s (massive open online courses), watching videos as and where I felt they would be necessary to grasp a concept (sometimes it is helpful for this to be explained).

However, I did not complete many MOOC’s.

Instead, I focussed on learning new skills and quickly applying them to data sets, or data science scenarios.

My point here is don’t get too hung up on completing entire courses for every skill, learn enough to do something practical, build things then learn why they work.

CompeteI found data science competitions to be a great resource for benchmarking my performance in particular for picking up machine learning.

There are two main sites where you can compete Kaggle and Analytics Vidhya.

Both have a wealth of data sets, the latter generally having more available for the beginner.

There are always open competitions with leaderboards where you can build models and submit them to find out where you rank.

This is particularly useful for learning to evaluate the performance of machine learning models.

It isn’t just about trying to climb higher up the board but more “is my interpretation of that model score correct?” and “how do I improve my model score?”.

Focus on soft skills tooTechnical skills are not the only thing that matter for a data scientist.

Data science can be hard you may have to explain a model to a non-technical colleague, convince a board to invest in a project, spend a lot of time cleaning data before getting to build an actual model.

You will need persistence and resilience.

Exceptional communication skills.


I worked on these skills, and am still working on them, alongside picking up the technical skills.

If you work to improve these qualities at the same time they will enable you to not only become a better data scientist, but also a better learner.

In the beginning, when I started to learn data science I spent quite a lot of time doing some of the don’t do’s I have mentioned.

For example, I spent time completing video courses in statistics and maths, thinking that I needed to do this before building any models.

Only for these concepts to not properly sink in until I was actually in the process of building something.

I have listed some tips to accelerate learning but in essence, these all equate to the same thing.

Learn enough to start building things, learn more to build better things, and repeat.


. More details

Leave a Reply