it’s possible to fully master every topic in data science?
With data science covering such a broad range of areas — statistics, programming, optimization, experimental design, data storytelling, generative AI, to name a few — I personally don’t think so.
Here’s a narrower question. Is it possible to fully master a single topic within data science? Sure, you can become an expert in some areas, but can you ever reach a point where there’s nothing left to learn? Again, I really don’t think so.
Every data scientist has something to learn, even those with extensive experience. The aim of my writing is to provide some insights from my learning journey that I hope will help you in yours.
This is the first part in a two-part series. In this article I’ll cover:
- Why you should continuously learn as a data scientist
- How to come up with topics to study
Let’s jump in!
1. Why continuously learn as a data scientist?
Continuous learners differentiate themselves
When I was younger, I studied Spanish in a group setting. Something interesting happened after the group became conversational. Many students stopped studying, they were content with their level of proficiency. Others continued to do daily study and practice.
At first, there wasn’t much difference between the two groups. But over time, those who continued learning pulled ahead. Their fluency, vocabulary, and confidence compounded, while the others plateaued.
Unfortunately, the same thing can happen to data scientists. Some stop learning after they have developed sufficient skills to do their jobs well. Similar to the Spanish cohort, early in a career, continuous learners and content data scientists will look similar. But as time passes, those who keep learning start to stand out. Their knowledge compounds, their judgment improves, and their ability to solve complex problems deepens.
Continuous learners and content data scientists will look similar early in their careers. But as time passes, those who keep learning will start to stand out.
Continuous learners shine because they can use their knowledge to come up with smarter solutions to problems. They will have a more mature understanding of data science tools and how to use them correctly in their work.
Learning brings fulfillment (for most)
This is a little bit fluffy, so I’ll keep it short. But I really do enjoy learning. I get a lot of fulfillment and satisfaction from taking some time to invest in myself and master new topics. If you like the idea of continuous learning, you will probably get a lot of fulfillment from it as well!
2. How to come up with things to study
We’ve established the value of career-long learning in the previous section, let’s talk about how to come up with things to study.
The best thing about studying on your own is that no one is telling you what to study. The worst thing about studying on your own is that no one is telling you what to study.
You’re not in school anymore, which is great. No more deadlines, no more exams and, perhaps most importantly, no more tuition. But you also lose the curated list of topics to study with corresponding materials, texts and lectures. Creating that is your job now! The flexibility of developing your own study plan is amazing. But the ambiguous, undirected space can be daunting.
Over the years, I’ve developed three approaches to come up with study subjects that work really well for me. My goal is that they can be a good starter for you to develop your own approach. Ultimately, you’ll have to find what works best for you.
Let’s get into the three approaches.
Topics from projects at work
If you are working as a data scientist, your projects will give you a rich supply of ‘deep dive’ study topics. This approach is pretty straight forward – study techniques/subjects that are pertinent to your work. Give special focus to areas where your understanding is the weakest.
For example, if you are designing an experiment, study experimental design. If you are solving an optimization problem, study optimization.
One great benefit of this approach is that it makes you better at your job immediately. You will have a deeper understanding of the problems you’re facing, and you’ll be able apply that understanding right away.
Following a “web” of topics
Data science is such a rich field of study, you can always go deeper on any given subject and so many topics are interrelated.
When studying, you’ll find many ‘tangent’ topics that are related to the topic at hand. I often take note of those topics and come back to them later. I call this the ‘web of topics.’ This is a great technique because you slowly build up a web of understanding around groups or related topics. This gives a deep knowledge that will differentiate you.
Here is an example of a small web of topics around logistic regression. I only included a few topics for the illustration – I’m sure you could come up with many more. Each one of the topics in the web have their own web, making a mega-web of related study topics.
I could keep going, but you get the point. Any individual topic will have a huge web of related topics. Keep a list of these somewhere and when you are done with the current subject you will always have a backlog of pertinent topics to dive into!
Note: Your web of topics needs to start somewhere. If you are having a hard time kicking it off, I recommend reading ‘The Elements of Statistical Learning’ or ‘Introduction to Statistical Learning’ by Hastie, Tibshirani and Friedman. These are foundational reads that will get you into a great web of study topics.
Discovery channels
Work projects and topic webs are two excellent approaches to curating a list of study subjects. However, these two approaches have a major blind spot. If you only use these techniques, you won’t be exposed to topics that don’t show up at work or in your natural sequence of study. There are likely really important topics that will be left untouched.
I use ‘discovery channels’ to help catch important topics that don’t come up organically. A discovery channel is any source of content that expose me to topics that are independent from my other studies. My main source of discovery channels are Towards Data Science, podcasts and YouTube channels.
When choosing a discovery channel, it is important to choose a source that covers a broad range of topics. If I, for example, followed a podcast that focused on experimental design – I probably wouldn’t source a wide array of topics to study from it. It might be a great resource for DOE study, but it wouldn’t be a good discovery channel.
I spend a relatively small percentage of my overall study effort on discovery channels, but they play the very important role in my studies.
Wrapping it up
I hope that this article leaves you feeling motivated to start independently studying if you aren’t already or has given you additional motivation to keep going if you already are studying. I also hope that I’ve given you a few fresh ideas on how to come up with things to study.
In a few weeks I’ll be posting part 2 of this article that will cover how to (1) avoid burnout, (2) choose learning strategies and (3) leverage solitude to cement and deepen your knowledge – stay tuned!
