A study from last year showed that there are nearly 100,000 positions in analytics and data science that remained unfilled because of a dearth of talent. Every day, new vacancies for data scientists are being posted to job sites, but good talent is still tough to find. “Though there is a huge demand for data scientists, there is a complete shortage of skilled data science professionals today,” asserts Brillio Data Scientist and Springboard mentor Indira Krishnamoorthi. This is primarily because traditional universities and educational institutions haven’t caught up to the needs of skills in future technologies. Therefore, aspirants are left to their own devices to gain skills in data science. “So, what do I do? Can I learn data science on my own?” is a question aspirants ask all the time. In this blog post, we compile the best advice for self-learning data science from Springboard’s mentors, who are leaders in the field in India.
Can I Learn Data Science on My Own?
“I was a below-average student at my college, with some interest in statistical analysis and data-driven predictions. When I joined Flipkart, even though I was a business analyst, I had the freedom to choose any means to build solutions for business problems. I wanted to use data science and machine learning. I did a lot of online courses. And I used this learning in my own projects to practice my skills,” says Abhishek Periwal, Data Scientist at Flipkart and Mentor at Springboard. If he can learn data science on his own, you can too!
Thomas Hepner did too. Hepner, who began his career as a financial analyst at Amazon, spent nearly 60 hours a week learning theory from online programs and practising it on Kaggle and other competitions. Today, he is a data science manager at Q2ebanking. If he can, you can too. With interest, discipline and persistence, you can learn data science on your own. Here are pointers from mentors at Springboard to get you started.
1. Taking data science courses online
The greatest gift the internet has given the world is easy access to information on practically anything; data science is no exception. There are plenty of data science courses — massive open online courses (MOOCs) — delivered by some of the world’s top data scientists. Many of them are free and offer largely similar curriculum and structure.
Pros of learning data science through MOOCs
- Most MOCs are free, while even the paid ones are largely inexpensive.
- They are taught by experts, both from the industry and academia.
- They offer global exposure, with use cases and applications from around the world.
Cons of learning through MOOCs
- There are too many options, which makes picking the right one tedious.
- Because they are generally taken by a large group, learners don’t get individual attention. Also, as they are recorded sessions, you will be missing the human element and motivation can dip. This combined with lack of support is one of the major reasons why completion rates are low in MOOCs. Think of it like joining a gym but with no trainer; it’s completely upto you to train yourself and keep yourself motivated.
- The curriculum is often theoretical and not geared towards job-readiness. In many MOOCs, there are no hands-projects, which restricts the amount of practical/technical exposure learners get. Even when there are assignments, the experience can be sub-optimal, and not resulting in jobs.
- Learners don’t get guidance for transitioning from the course to a job. In fact, a study by Stanford points out that students who worked in teams were 16 times as likely to pass the course and just adding a mentor increased the sign-ins to a course by 13%. This is often not available in MOOCs.
2. Reading books and other online resources
Another time-tested method to learn anything is through books. There are plenty of books published about data science, whether you need a general overview or specific analysis. In fact, you can also start learning the basics of data science such as mathematics, statistics, programming, etc. before jumping into the field.
Here are some books our SME mentors recommend:
- Python Data Science Handbook by Jake VanderPlas: As the name suggests, this book is a primer on statistical and computational skills needed to practice data science with Python. It offers a good foundation for the subject.
- An Introduction to Statistical Learning: with Applications in R: This book delves deep into the theory and fundamentals of statistics needed to build data science careers.
- Doing Data Science: Straight Talk from the Frontline by Cathy O’Neil and Rachel Schutt: This is a practical guide, which needs you to know linear algebra, statistics and programming to use it.
If you’re a books person, here’s our top 10 data science books for you.
Pros of learning data science from books
- You have plenty of options, you can pick exactly what you need.
- You have easy access — many books are available as e-books and audiobooks as well.
- You can go back and refer to them as many times as you need.
Cons of learning data science from books
- You have plenty of options, it can be difficult to decide when to stop reading and start practice.
- Studying from books is a lonely pursuit. As Jaidev Deshpande, Senior Data Scientist at Gramener, suggests, it’s important for every learner to have someone to talk to — which is not possible with books.
- It can be time-consuming, reading a book might take a week, while watching videos in an online program might be quicker.
3. Topping up with practical experience
As a highly practical field, data science can not be mastered with MOOCs and books alone. If you’re looking to build a career in data science, you need to demonstrate that you can do data science, not just know it. Hackathons and competitions can help you achieve that. A good place to start is Kaggle, a platform that offers hundreds of data science competitions and projects. Driven data conduct data science competitions for social good. Machine Hack lists some of the toughest challenges out there. It also helps to set up a GitHub profile — to contribute to open source projects, showcase your work, collect feedback and collaborate with the best minds in the field.
Pros of gaining experience from competitions
- You get access to real-world datasets and business challenges.
- Sense of community competing against peers from different parts of the world.
- Ability to learn from winning solutions.
Cons of relying on data science competitions
- It can get lonely and frustrating when you’re working on your own.
- Many of these competitions are difficult to win, which can sometimes be demoralising.
- Datasets in these competitions can be too clean / pre-processed, making it distant from the real-world datasets you work within data science careers.
- The environment is certainly not a perfect alternative for real-world work experience.
4. Experience through internships
An internship is the closest thing to a real job you can get in data science. It can help you learn from professional data scientists about their everyday tasks and responsibilities. It can help you to widen your network that plays a crucial role in your job hunt.
Remember that you can’t learn everything in an internship. It is often the last step in your learning journey, smoothening a transition to a full-time career. So, to land an internship in data science, you need to already have data science skills. Look at this 36-week internship at Target for instance. It expects prior work experience, in addition to skills.
On the other hand, this internship from CodeMonk doesn’t ask for experience, but data science skills are required.
Pros of internships
- Experience in real-world projects.
- A step in the door of a data science career.
Cons of internships
- Unless it’s a clearly designed internship geared towards making the most of your skills, it can be unhelpful and a waste of time.
- Depending on the length of the internship, your learning might be limited.
- An internship is no guarantee that you’ll get a job at the end of it.
How Do I Learn Data Science?
“One drawback of being self-taught is that you will have “holes” in your knowledge, that you are not aware of, but they will show up one day, usually at the worst time.”— Vincent Granville, Data Science Executive
In order to learn data science without such gaps, you need more than just ‘resources’. You need a curriculum that is aimed at job-readiness. Lectures, readings and assignments need to prepare you for a career. You need at least two portfolio projects, which solve real-world problems and are evaluated by experts. Like Jaidev suggests, you need someone to talk to — a mentor who has walked your path and can guide you through the ups and downs in your journey. It needs to offer a certification, which holds value among potential recruiters in the market. The online course you choose needs not only to teach you data science but also enable your transition to a data science career.
This is what Springboard’s data science career track endeavours to do. In addition to a job-oriented curriculum and hands-on projects, it also includes 1:1 mentorship from some of the best minds in data science today. Through weekly calls and regular catch-ups, they will answer all your questions and prevent you from making the mistakes they did, and help you apply concepts to real-world projects, accelerating your data science career transition. The career track program is project-driven and you will also get a career coach to help transition to a job in data science. Oh, and did we mention the job guarantee? Apply Now!