University of california berkeley masters in data science

How I chose this master’s program in data science over MOOCs, certificates, and other universities, and whether it’s worth it.

Someone reached out to me this week to ask for my perspective on UC Berkeley’s Master of Information and Data Science (MIDS) program. It’s a part-time online master’s program that I’ve been actively pursuing for the past year. “Overall I’m interested in if you’d recommend it in terms of price vs value,” he asked, “and if you explored other programs how did you end up choosing MIDS?”

(Also see my update two years into the program: “How a Part-Time Data Science Master’s Program Changed My Life”)

“Overall I’m interested in if you’d recommend it in terms of price vs value — and if you explored other programs how did you end up choosing MIDS?”

Two years ago, as I was evaluating various data science program options, including Berkeley, I had the same question. Why should anyone pay the higher cost to pursue a formal master’s program in the age of MOOCs and self-learning? Among master’s programs, Berkeley isn’t the most affordable either. The University of Illinois, Urbana-Champagn (UIUC) offers a similar program for just $20,000, whereas Berkeley’s program costs closer to $70,000. So is it justifiable? When I searched for the answer online, I could only find some outdated Quora posts on the matter with less-than-helpful information. I did a quick search today, and I see the same posts. So I thought I’d give you an insider scoop, current as of summer 2020.

This post aims to describe why I chose the program and why it was worth it for me, as a current student who completed a little over half of the program. Whether the program is valuable to you would depend on your personal situation and goals, but in sharing mine, I hope to make your decision easier.

I started as a data analyst. Photo by Carlos Muza on Unsplash

Why I Wanted to Learn Data Science

A little about me: my career has always been at the intersection of business judgment and data analytics. As one of the few people proficient in SQL at KPMG’s Risk Advisory practice and Uber’s Internal Audit team (in Finance and Accounting), I translated legalese into queries, crystalizing stories from data that supported audit recommendations. I used Benford’s law, keyword searches, and basic statistics to identify fraud schemes and other risks, and I took pride in solving complex puzzles with data.

And yet, I found that merely diagnosing past events was neither efficient in the long term nor intellectually satisfying. By 2017, I had tasted the additional flexibility that learning basic Python had brought into my workflow, allowing me to expand beyond relational databases. I witnessed my incredibly talented colleague, Daniel Piers, offer a critical piece of evidence for an investigation using an ML algorithm. The industry of risk management, which had traditionally focused on descriptive or diagnostic analytics, was adapting more predictive and prescriptive analytics. The latest fraud monitoring tools leveraged machine learning to detect forged documents or dubious expense claims. And I wanted to be at the forefront of this revolution.

I started training myself in data science through a MOOC, but soon realized its limits. First, I had to build my own curriculum, but the issue was that I didn’t know what I didn’t know. I also had a hard time finding high-quality contents to map to my learning plan, since I didn’t know how to evaluate them. The courses I tried tended to focus on tools, whereas I needed a little more help with developing a data science mindset. And lastly, I didn’t have enough incentive or pressure to keep myself accountable and finish the courses I started. So I decided to look for alternatives — data science master programs.

I had some decisions to make. Photo by Letizia Bordoni on Unsplash

What I Needed From a Data Science Program

A Legitimate Master’s Degree with a Brand Value and Credibility

If I was going to dedicate time and money, it wasn’t enough for me to learn something, but that other people would be able to tell that I learned something. I considered shorter and inexpensive DS certificate programs, but I didn’t know how much credibility they would lend to my resume. So I decided to focus on master’s degree programs from well-established higher education institutions.

Part-time and Remote

I was on a great career trajectory at Uber, and I wasn’t about to give that up. While developing DS skills was important, so was developing leadership skills and domain expertise. A part-time program would allow me to keep my income, continue my current career trajectory, and pursue further education.

I didn’t care where the school was, as long as I could access it from my home in San Francisco. This ruled out the University of Chicago and the University of Washington, which had part-time programs but only on campus.

Lower Barrier to Entry for Technical Skills, but Opportunity to Learn

I was a business major who knew a thing or two about analytics. Sure, I had written a few for loops in my life, but I couldn’t comment on any data structures and algorithms. The last formal education I received in statistics was in junior year of college, and the last one in mathematics, IB Calculus in high school. Unfortunately for me, several programs required the applicants to be comfortable with linear algebra. I guess I could have studied it on my own from LSU or something, but I was itching to get started, so I ruled those programs out quickly. These included UIUC, UCLA, and Johns Hopkins.

I also did not want to veer too far into the other end, which was business analytics. I already knew business analytics. So I ruled out programs such as those of Columbia (in-person and in San Francisco, but it was also $80K!) and Indiana.

Network and Community

My philosophy is that I can’t possibly learn everything there is to know, but I can always call friends. Therefore, I wanted a program in which I could network with students and instructors in a meaningful way. This is the main reason why many of the MOOCs did not work for me.

What attracted me to Berkeley was its community. The MIDS 2017 Status Report said:

In 2016, a program survey indicated that 72% of students agreed with the statement, “I feel like a member of my university community” and 83% of students agreed with the statement, “This program has helped me develop a network with fellow students.”

I was sold.

Having spent the last year in it, I completely agree — remote learning could not stop me from fully integrating into the community. The small live classes (<=15 students), frequent office hours with faculty, active Slack channels on various topics, and occasional local meet-ups all add up to the sense of belonging.

The Final Choice

According to my spreadsheet, there were only three schools that met all of my conditions: Notre Dame, Northwestern, and Berkeley. Since I had completed my undergrad in Berkeley and loved it, it automatically scored higher in my heart. I also knew that if there were any on-campus events, I could easily visit from San Francisco. So I decided to apply to Berkeley first and keep the others in my back pocket. And I was accepted.

What was I supposed do? Stop investing in my career? Didn’t think so. Photo by Avel Chuklanov on Unsplash

The Berkeley Experience

After I got the admission letter in March 2019, I still wasn’t sure. I had gotten only a small scholarship to barely make a dent in the tuition and needed to pay the remaining out of pocket, so I hesitated. I spent many hours browsing Quora and reaching out to current or former MIDS students on LinkedIn. Data science seemed like something I could study on my own on YouTube or Coursera, and as long as I could build a strong portfolio, I would be able to develop my career in it. But then I reflected upon my journey the last couple of years, on how difficult it was to determine what I even had to know. I needed guidance, and I needed a community. So I decided to sign up. “I can always drop it,” I thought. Worst case, I would have spent a semester networking with the students of the nation’s top public university. I’d walk away with a good understanding of the program’s curriculum, so I’d know what to study on my own next.

Three semesters later, I couldn’t be happier with my experience. I started applying what I learned at school to my job in the first few weeks of the program. Last summer, I was struggling to explain what a confidence interval was, but by now, I have written two minor research papers in statistics. Today, I can train ML models, explain the inner workings of them, and apply hyperparameter tuning or dimensionality reductions. The most significant sign of my growth, however, is in my aspirations. In the past, I wanted to become a business professional well-versed in data science, but now, I wish to become a data scientist well-versed in business. I went into the program thinking, if I can just talk to ML engineers intelligently about the concepts and discuss how to use it for business, that would be enough. Today, I’m considering a career more deeply involved in ML algorithms.

The Academics

Berkeley did not have a hard requirement for math or programming at the time of application but offered bridge courses and support for establishing a strong foundation to build on. Berkeley’s ability to help bridge this gap is an incredible asset and an advantage to the program. It allowed business folks like me to bring interesting perspectives to the table so that more technical people could learn from us and vice versa. The MIDS program is excellent at managing diversity and inclusion of thoughts, disciplines, and backgrounds.

The quality of the contents is high — that goes without saying. The courses address the fundamentals that never go out of style plus the latest data science trends. The top-notch instructors, who range from professors to top data scientists from large companies, are continuously innovating the program to stay a step ahead of the game.

Unlike a MOOC, Cal builds accountability into the program, making it easy for me to dedicate 20–30 hrs/week toward learning. The live sessions with only 15 students motivate me to finish all the weekly materials in time so that I can contribute to the discussion. The live sessions and office hours give me a chance to ask questions, which is the third-best way I learn. The second best way is by doing, which is taken care of by hands-on projects. The best way I learn is by teaching, and by signing up to be a TA in a programming class, I got to do this a little bit, too — not exactly by teaching, but by providing feedback on other students’ code. In case these aren’t enough, my dedicated student advisor would check in with me regularly and ensure I have everything I need to succeed in MIDS.

The Community

The School of Information also offers an excellent career counseling service. I’ve already had two private advising sessions with the fantastic Laurie Haskell-Woerner — our career counselor — and benefited greatly from her insight. We’ve discussed some short-term plans like improving my LinkedIn presence and long-term career strategies. She connected me with several alumni who could advise me further on specific topics. To receive such a tailored and high-quality care felt like having a personal career coach and mentor.

Last but not the least, the community engagement has exceeded my expectations. I’ve made many friends since there were ample opportunities to speak to people face-to-face and get to know them. I love the student-led study sessions. I appreciate the diversity of students from all walks of life: engineers, statisticians, teachers, finance leaders, entrepreneurs, and architects. Slack enhances the experience. One of the first things I did on Slack was to ask for some career advice, and so many students jumped in to help. I also love #coffee, #music, #cool-data-viz, #ethics, and #no-stupid-questions. Students and faculty from around the world, across all industries, age groups, and walks of life share their perspectives. It’s quite beautiful.

Time, money, and hard work? Yes. Transformational? Also, yes. Photo by Suzanne D. Williams on Unsplash

So, Is It Worth It?

Yes.

I am glad I did not let the price tag hold me back from pursuing my dreams and pivoting my career. I don’t know you personally, dear Reader, and can’t speak to whether it will be as fruitful an endeavor for you as it has been for me. However, I see a diverse group of intelligent people who are enjoying the program. There seems to be something for everyone in MIDS. Someone with a Ph.D. or a splendid career in business; someone who’s won Kaggle competitions before or someone who’s just starting to dabble in scikit-learn; and someone reaching their retirement age or someone who just graduated college. Everyone’s goals and journeys are different from mine, but there’s something that works for them here at Berkeley.

If you are reading this because you are on the fence about MIDS, I say, go for it! If you had any doubts about online graduate programs before, keep in mind that MIDS has been investing in the remote learning technology long before it was cool to do so, since the pre-pandemic era. And yes, online students get the same treatment and benefits as any other Berkeley student— the libraries, the student discounts, the right to vote for the student officials, you name it. I even made my way to Berkeley to pick up a student ID to show at museums and use as a free bus pass on AC Transit.

The program has offered me much more than I bargained for, and I am grateful for it. My only regret is not having started it earlier in my life.

(Note: Upon further research, I found this great new post by Wei Wang on Towards Data Science that provides a program overview and course reviews, written in April 2020. Check it out!)

(Update: Check out this new article called “1 Year Into the Program: Georgia Tech OMSA vs UC Berkeley MIDS” by my classmate and his friend, too. Written in June 2020.)