Given the hype going on about data science, this is a very valid question, do you need a master’s degree. If this hype is genuine or not is also another big question. But this article will focus on if a master’s degree is necessary. There are so many other short, easier, and cheaper options out there. In fact so many free courses in Coursera, edx, and learning sites. Is it still necessary to go through a big academic process? Assignments, quizzes, exams, term projects, presentations?

This article will cover:

1. A little bit about my own journey
2. How I feel about being a master’s student currently.
3. Do I still believe that one can be a successful data scientist without a degree?
4. Some hurdles I faced when I was a self-learner.
5. Some suggestion to overcome the problems of self-learning

A Little Bit About my Own Journey Before I Started my Masters

I myself had this debate in my head for about a year before I started my master’s. My Bachelor is in Civil Engineering and my Master’s in Environmental Engineering. I did not write a single line of code during all these years in college. I started learning to code after my master’s when I was a teacher. I got a free six months Bootcamp through LaunchCode.

Like most beginner Bootcamp, it taught me the basics. I took so many courses from Coursera, Udacity, and Udemy on data science, statistics, python, and machine learning. I wrote in details about those courses and how I switched to data science in this article.

Luckily I was able to do two internships and a few freelance projects. I realized I have a lot to learn. But I couldn’t really figure out where to find the right materials. What to learn next?

When I took the statistics course in Coursera, it slowly became too hard for me. It was a three-course specialization. The first two courses were not bad but the third one was really hard. I thought I needed more help. Also before I took this specialization I was taking another one for some time. Soon I realized that was not the right one for me. Some waste of time there.

I took some machine learning courses as well. There also took one specialization for some time and then realized this is not the one for me. Then I found professor Andrew Ngs Machine learning course. He is the best person I found who can break down the machine learning concepts into pieces really well. By the way, I took those courses for free. There is an audit option in Coursera. You can audit a course as many times as you want for free. I did not spend any money there.

I also took some SQL courses and deep learning courses with Tensorflow. That made me feel pretty confident that I learned a lot. But still, when I saw the requirements in job postings I hardly found something that matches my knowledge. Sometimes I pushed myself to learn new skills for a few months because I saw them in job postings. At a point, I realized I will never meet all the requirements in all the job postings. I need to choose some skills. Now the challenge was to figure out what are those skills?

At the same time, I was learning all by myself, alone at home all day every day. There was no one to talk to when I am stuck. At times it was tiring.

Though I wanted to avoid going for a master’s because I already have a master’s. But I realized maybe it is necessary to take a well-organized master’s degree. That might give me confidence in my knowledge, plus acceptance and recognition to other people. Though I know many people who do not have any degrees in data science or even computer science but successful in the field. But I thought I probably am not that smart or not that lucky!

I need to add, you may ask why a master’s degree, there are a lot of boot camps that look so good and promising! I just could not find one that I could really believe in. I looked into so many. A lot of them are six months-long and their curriculum is so big! They did not look realistic for me. I felt like they would probably go through the basics that I already learned. They may be a good starting point for a complete beginner.

Again, I found one that promises to help you find a job. The teaches each topic for three weeks. Like programming three weeks, statistics three weeks, data mining three weeks, machine learning three weeks, big data three weeks. But after going through all these specializations and courses already I at least knew this much that none of this topic can be covered in three weeks. For programming, machine learning, statistics, even 12 weeks is too little. It took me almost 6 months to finish Andre Ngs’s machine learning course. That’s a beginner level. So I thought those boot camps will teach the basics that I already have. I probably should go for a master’s that will teach me something more than basics. But probably for a complete beginner, a Bootcamp is helpful if money is not a constraint.

So, I should say, self-learning helped me make decisions.

How I Feel About Going Through a Masters Program in Boston University?

To be honest, great! I genuinely believe it was an excellent decision for me. Here is the breakdown. I learned the programming basics and some beginner-level algorithms in the boot camp. But they are really the starters. Later on, Coursera taught me some algorithms. Those are also good but in this master’s program, I learned about some advanced level searching and sorting algorithms, dynamic programming, memorization, and much more. I also got an idea about how much I needed to learn by myself. Because a few weeks or 2/3 months of programming courses are never enough to make you an expert in algorithms unless you are a superman or superwoman. It takes a lot of practice to become really good! But at least you need a clear picture about what to learn if you are coming from a totally non-CS background.

Then two terms of data analytics courses that are mostly on statistics in R. I kept thinking of learning R, now I had to because of this course. As I mentioned before, I took a specialization in statistics which was pretty hard and I thought quite advanced. In the statistics courses of a total of 12 weeks, I found so many topics that I already covered before but missed so many pieces that were small but very important. I did not realize it simply because there was nobody to guide me when I learned by myself. At the same time, I learned so many advanced topics that I did not know existed. Also, I could not learn this much material in 12 weeks by myself.

There are some courses like data mining and big data. I thought I already learned the big libraries for computation and data manipulation like Numpy and Pandas, Data visualization libraries like Matplotlib, Seaborn, Machine learning, and deep learning using python from scratch and also using inbuilt libraries like scikit-learn and Tensorflow, I learned SQL. I am just done and ready to rock and roll. I do not need data mining and big data at all. But now I am learning those as part of my master’s program as well.

So I am really happy that I made this decision. Do not get me wrong though. There will be a lot of self-learning even if you finish a master’s degree. You just will have a strong guideline on where to take it forward after this.

Do I Still Believe That You Can be Successful Data Scientist Without a Masters Degree?

Yes. Because I do not have a choice. I know so many people doing great without a master’s degree in data science. They came from a totally different field.

It is never been easier to learn by yourself. Because so much material is out there that you can take advantage of. Here I made a list of high-quality free resources that can be used to become an expert in the field.

If you are interested in deep learning and machine and learn very well, here are a some great free resources.

In fact, the algorithms, statistics, data mining, big data that I mentioned before are all out there. Great quality free courses, books, so many blogs, youtube videos are available to learn. And many big-name companies like even Instagram and Google hire people with knowledge even if you do not have the degrees.

Then why spending money and take the hassle of a master’s degree, exams, assignments, and all that?

Mistakes I made while learning by myself:

Yes, everything is out there. In a way that is good and bad at the same time.

1. Because you have so many options, it creates confusion. If you ask some experienced people, you will get so much advice and opinions that will make you more confused. Because different people may show you a different route. Where to start, what should be the sequence of learning, what else to learn, how much to practice, too many questions. But no one to answer them.

2. When I took free courses in Coursera or edx, I was trying to finish everything fast so that I can enrich my resume and can start working fast. Now I realized I needed to spend more time and needed to allow myself to digest the materials well.

3. Choosing the right course becomes a challenge. A lot of time I started taking a course and after a week or two, I realized that it was not appropriate for me. So some waste of time there. That will always happen for self-learners. You have to accept that. Also before jumping into something that you heard is good, you need to take some time and check it well.

4. It’s hard to find good practice materials. So, I learned the ideas but probably did not practice enough.

5. Most of the time I focused on the learning of the coding part. But data science is not about coding only. You need to see the problem, render the question, and then code up the solution. For that, the necessary thing is an analytical ability, some theoretical knowledge, the ability to see the data gap, and an understanding of the subject matter so that you can come up with the question.

6. Lastly, a lot of people will agree with me on this. You may start with a lot of enthusiasm but after some time most people have a hard time finding motivation. Because it’s not about 2/3 months only. It’s a long-term commitment. Though there are a lot of boot camps out there that promise to make you a data scientist in 3/4 months. I have an article on that.

Some Suggestion to Overcome the Problems of Self-Learning

1. If you can find a partner who has the same goal as yours or a group is even better. That is the best for you. That will be very hard to find. But if you are lucky enough to find some people to study together, discuss problems when you are stuck, that will make things a lot easier.
2. Do not rush or try shortcuts. That never helps. Allow enough time to learn to program, statistics, each concept or library very well. It can be alluring to move to a new thing, learn a new programming language before learning one language well or move to new libraries. But that will only waste your time. When you finish learning one thing, you need to ask yourself, are you confident that you can use it well by yourself in a project?
3. Fighting distraction is a huge work. Writing helps a lot in that case. I work all alone at home. It is very easy to get distracted. Youtube, Google, Facebook, Twitter, Instagram so many distractions. Especially I am a bit addicted to YouTube. I think I could control it a lot now. Still, a lot of work is needed to totally get out of that addiction. I found one thing very helpful. Keep reminding myself throughout the day what I have done today till now.
4. One way to fight distraction and stay motivated is to say it loud. If you have a study group make a commitment to them or say in loud what is your goal for the next three days or tomorrow. Or you can do it to your family members. But they may get bored in a few days. Keeping a journal helps if there is no one to talk out loud.
5. Keep making short-term goals. Make plans for the day or two days. Most of the time when we make a long-term goal, we become complicit. Most students study at the last moment for an exam though they had a week to study.
6. Go to meetups, conferences, seminars, and meet people, have a conversation. Try to know what they do in their jobs, their backgrounds. That will give a lot of perspectives and open your eyes to many new things. Slowly you will make a community of data professionals. That also a great way of staying motivated and keep working on it even if the success looks far away. This is helpful for people of any stage actually.
7. Watch other people’s presentations carefully. How they approach a problem. If you are a beginner, Kaggle can be a great resource. It has some great free courses as well as a lot of example projects. With free datasets, it has example projects and workouts with those datasets. If you are just planning to make a portfolio, you can draw a lot of inspiration and ideas from those projects.
8. Social media can be very helpful. I know there is a lot of negative notion about social media. But I find it very helpful. Follow the right people on Twitter and Instagram, connect with professionals on LinkedIn, and join different good Facebook groups. There are Facebook r groups for very expert professionals as well as for learners. Find the ones that suit you. It’s a great way to find information, community, and motivation.
9. If you are lucky enough to find a mentor, that will take you a long way in a short time. It will save you from losing a lot of time. I heard it from people a lot. But I was never lucky enough. I never had a mentor.

Conclusion

You may find it a bit confusing. Actually, it is all about your personality. There are people who can work and study well by themselves. There are people who need to go through an academic process. But becoming a data scientist is a long journey if you are a complete beginner. Even if you do a master’s there will be a lot of self-learning. I hope this article was helpful for you in making your mind about a master’s degree or moving forward. I wanted to share my experience.