I have been sitting in my room, by the window, staring at my laptop from 9 AM to 5 PM every day the past three months doing a Data Science Immersive Program offered by General Assembly (GA). What it is like doing the program? In short. It’s a lot — a lot of stress, a lot of learning, and a lot of frustrating moments. I even questioned myself several times “Why am I doing this to myself?”
It’s challenging and it’s all good. Today is the last day of the program. While it’s still fresh, I would like to document and share my learning experience with people who are interested in doing this data science program.
What this GA Data Science Immersive Program is about?
It’s like a tunnel. You walk into the tunnel as somebody who have little data science background and come out of the other side of the tunnel feeling confident as a data scientist. It sounds magic. Right? The truth is — there is no magic. It’s a lot of hard work. Period.
Who are the people taking this program? Can anybody do it?
Most of the people doing this program are people genuinely interested or passionate about working with data. Some people have more background than others, and there are also people like me, who have very little technical background.
Starting from the very beginning of the whole process, if you are interested in the program and contact GA, they will reach out to you, talk to you to understand why you want to do this, what it is you want to achieve, and whether you seriously commit to it. If your intention is confirmed, they will give you a data project to do. At this stage, it doesn’t matter what tool you use. You work on the data and present your finding to GA. They will then come back to you, invite you to enroll to the program, or not. It’s actually quite a thorough process, including a job searching survey to make sure there are jobs out there you would like to do after graduating from this program.
What is the content and class experience?
In terms of tool we use, this program introduces data science using python and a lot of its libraries. We started with basic python skills, pandas, and numpy as tools for exploratory data analysis (EDA), and Matplotlib and Seaborn for data visualization. With machine learning, we used a lot of libraries offered by Scikit-learn for model configuration, feature engineering, model regulation, train-test configuration, and so on. This is just the basic.
There was also TensorFlow for neural network, aka deep learning, and Natural Language Toolkit for natural language processing. In the class we also took a quick look at different BERT, Word2Vec, and GloVe (for natural language processing too). We learned how image classification works. Not to say, later on there were also time serious data modeling and Bayesian modeling. There were a lot more. The list is long. It’s 480 hours of learning condensed in 12 weeks. In terms of college courses, it’s like taking twelve 3-credit courses within twelve weeks.
On top of the tools and developing technical skills and familiarity, there are also a lot of math. Since machine learning is to apply statistic modeling to data utilizing computer power, understand the algorithm and how it works is a critical part embedded in every new model we learn. To get a good model, we also need to understand how to evaluate the performance through reading and understanding metrics. It is math heavy too.
Ok. So, it’s a lot. How well a learning process can be in such an intensive format?
To begin with, although it’s remote, all lectures are live and GA really put in a lot of effort to make sure it works for everybody. The instructors are great with teaching and communication and the TAs are always there, always available to help.
In the class, there is no sit around and watch — it is designed for everybody to code along through the whole program. In the classroom, while the instructor is going through the class material, there are TA accompanying every class — in case anybody needs technical support. So, we are actually learning data science by doing it. At the same time, it’s also a great classroom experience learning from the instructors’ coding and data modeling styles.
Other than the classroom material and experience, what else?
Well, on top of the regular 9–5 class hours, there are also twenty labs, six quizzes, and six projects included in the program. Labs are designed as a compliment material to what we learned in the class. Doing lab is the time for us to be on our own feet, tackling data science problems ourselves, practicing using tools we just learned. Besides labs, projects are like test stones to tie up tools learned from different stages together. The purpose of quiz is to simulate interviews for us to get use to coding or answering questions under stress with time limits.
I can go on and on about this program and what is in it. In short, it’s a lot. No kidding. There are so many materials that I felt it’s actually like a mini computer science program, only it has a focus on data science. In terms of how much a person can learn, it depends. Everybody’s learning style is different, and the determining factor is how much time and effort a person would invest and how much pain a person would endure.
Life as a GA, DS student
First, this is my experience, doesn’t mean everybody’s life is like this. I normally woke up 5 or 6 o’clock in the morning thinking about python or pandas, my labs, class materials to review, stuff to do for my projects, presentations, and so on. I normally started at 6 AM-ish, working on what woke me up that morning, and got ready for the first session started at 9.
Starting from 9 o’clock, there were back to back zoom sessions with one lunch break and a few coffee breaks in between. Well, time certainly went by fast. Before I realized it, the 9–5 regular hours were over. After grabbing a quick meal, I normally went right back to my room, working on the lab, projects, or reviewing materials till seven o’clock-ish, taking a break, and then kept working till I was so tired that I couldn’t look at the monitor anymore. There was never shortage of works to do. How about the weekend? Weekends were normally the time for me to work on class stuff that I didn’t have time to do during weekdays. A lot of times, I finally caught up the material of that week over the weekend, feeling good for a few hours and then feeling I was falling behind again after classes on Monday. My life was like this the first eight weeks.
I don’t think it’s like this for everybody. It’s tough for me because I had very little programming background and skill, I was not a big fence of math, and I knew nothing about machine learning. I struggled a lot to keep up with the class and constantly worrying about falling off the wagon. I mean, there is no guarantee of pass or completion. My goal for the first few weeks was to not fail the program.
How does it feel now that the program is completed?
It feels GREAT!! It doesn’t feel real though. I can’t believe it’s done. I am so happy that I didn’t quit and I didn’t fail. Does my life all of a sudden full of sunshine? No, it is not like that. I have learned a lot and the science part of the program also changed the way I think a little. I am still me, but I can feel the difference. I can also see opportunities that I couldn’t see or reach before. I achieved my goal to built my own data science tool box for my next adventure.
My goal is never to become a data scientist — two third into the program I did consider going after data scientist jobs but I soon realized that it is not for me. Coming out of this tunnel, I am now a data-science-literate person and I also develop new insights about myself. Can I do more and learn more? Oh, yeah. No doubt. This is just a beginning.
Text or email me if you are interested in data science and want to know more about it or the program.