Aravind is no stranger to statistics and analytics - he has a Master’s degree in Statistics from Columbia and has been working as an Analyst at a global investment firm building statistical models. But Aravind wanted to strengthen his programming and machine learning skills, so he considered his options and chose NYC Data Science Academy to take his skillset to the next level. Aravind candidly answers all of our questions about why he chose a data science bootcamp over a second Master’s degree, his final projects, and how data science skills have made him a better analyst.
What were you up to before you decided to go to NYC Data Science Academy?
I didn’t attend NYC Data Science as a typical “career switcher.” Instead, I was mostly interested in gaining new skillsets quickly. Bootcamps offer an intense curriculum, but at the same time, are shorter than traditional options.
I already have a background in statistics and have been working for an investment firm as an analyst. I worked with different groups at the firm doing statistical modeling, but I didn’t have as much machine learning and programming experience. That’s what drove me to NYC Data Science Academy.
So you wanted to move up in your career, not change your career?
Yes. I could have continued as an analyst, but data science is a skillset that is designed to solve real world problems using data driven methods. It requires a strong understanding and domain knowledge of programming and statistics, and that was my goal.
Once you decided that you wanted to learn those programming and machine learning skills, how did you research your options?
I had two other options, which weren’t bad, but had their downfalls in that they were time-consuming and expensive. First, I could use online courses. The content in machine learning courses on Coursera is very good, but it can take over 8 months to complete a set of courses.
I also considered doing another Master’s degree, but I would be out of work for a long time, and about 30 to 40% of the coursework would overlap with my Statistics Master’s degree. So I decided that a bootcamp was the best option.
How did you decide between NYC Data Science Academy and other data science bootcamps in New York?
I was already familiar with Vivian Zhang’s teaching from her meetup groups, even before she started NYC Data Science Academy. I had applied for the first cohort, but it started in early 2015 and I decided to postpone it for work commitments.
I looked at both The Data Incubator and Metis. The Data Incubator didn’t have a class that started immediately, which I needed. I looked at the coursework at Metis, but they primarily teach Python, and I wanted to learn both R and Python.
I chose the Data Science Academy because of the variety of coursework they offer. We used both R and Python in great detail. Both languages are useful for a data scientist; neither is “better” than the other. I feel that R, for example, may be a great data visualization tool, while Python could be used for analytics and machine learning. At the same time, the latest machine learning packages in R have been promising. Getting exposed to both R and Python was appealing.
At NYC Data Science Academy, were you satisfied with the emphasis on those programming and machine learning skills that you wanted to learn?
There was plenty of material in the curriculum, but we also had a lot of coding sessions where we could sharpen our coding skills. If you really want to become a better programmer, then there is a lot of work that you have to do on your own.
Tell us about the projects that you created while you were at NYC Data Science Academy.
We worked on five projects throughout the camp. We had to complete projects and do presentations, then start on the next project immediately. We were always able to complete those projects in the designated amount of time, but it was very intense.
The projects that we worked on for data visualization were individual projects. The machine learning and capstone projects were group projects.
Can you tell us about your Capstone Project?
My capstone project involved the classification of musical scales. Earlier studies show that songs in different genres can be classified based on signal information. We used classification algorithms to decide whether a particular scale is rock, hip hop, etc. Sometimes you can even classify based on characteristics like whether it’s a minor or major key. Or even more specifically, the mood of the music. For example, there’s a concept called “raga” in Indian classical music with a specific frequency pattern. We fed the computer existing data with what we know about raga, then built a system that automatically classifies music. Companies like Soundhound do a lot of this fingerprinting, which involves a lot of machine learning and digital signal processing.
My Python project was to build a web scraper to collect and analyse rental listings on Streeteasy.
Who was your instructor at NYC Data Science?
Our primary instructor was Christopher, who came from a statistics background. I thought he did an excellent job teaching and communicating each of the algorithms and statistical concepts. He was clear, concise, and effective.
You have a Masters degree in Statistics and have been working with Statistics for the last few years- do you feel like you still learned a lot from Christopher?
I wouldn’t call myself a statistics expert! Even the way Christopher approached simple concepts was interesting. Often with stats, people approach a problem without understanding the conceptual underpinnings behind a particular idea. Chris was able to explain both the mathematical concepts and the conceptual underpinnings.
For example, conceptually, we may say that a t Distribution is “fat tailed” compared to a normal distribution, but Chris would explain why this is so instead of making those basic assumptions.
Did the rest of your cohort have the same background as you? Were there people with different levels of education?
One thing I learned is that at a bootcamp, everyone comes from varied backgrounds. Some students had a Master’s degree in a non-quantitative subject, others had Bachelor’s degrees. Some even had math and physics PhDs- and among those PhDs, some had a theoretical background, while others had programming experience. Those with a computer science background had a small advantage because they had less catch up to do for programming prework.
Everyone had an area that they wanted to improve on. I came from a statistics background, so I was able to focus on topics that I hadn’t had a chance to work on before, like Python.
What was the biggest challenge you faced at NYC Data Science Academy?
During the bootcamp, a bout of flu went around! I had to miss a couple of classes, and then quickly complete a project and present it. I wanted to ensure that the quality of my work didn’t suffer, so I had to work extra hard. I wasn’t sure I could do it, but the support of the TAs was so helpful. Chris made the lessons that I missed available on video. All of those things helped me bounce back and complete two projects really quickly.
What are you up to after graduating in March?
I am with the Asian Markets strategy group that tries to use both qualitative and quantitative strategies for Indian and Chinese Equities. My idea is to contribute to quantitative groups at the company in a better way through machine learning and automation of processes.
Have you gotten to put your new skills to work?
With my programming skills, I’ve been building a tool that takes information from the web about particular news articles about stock. I’m using natural language processing to use that news information in a more seamless way. Plus, my supervisor also feels like those quantitative skills are helping the group.
Were you impressed with the feedback loop at NYC Data Science Academy?
One of the things I have to mention is that Vivian is doing a great job keeping the best aspects of each cohort, and at the same time making sure that each cohort is better than the one before. The feedback mechanism that exists at NYC Data Science Academy is really impressive. I was surprised at the extent to which Vivian valued my opinion as a graduate.
In this new cohort, I made a couple of suggestions, and they have additional hours dedicated to MongoDB and they’re working on a machine learning “defense exam,” which would go with the final project and would be very useful for someone who wanted to prepare for a job. They would get experience with theory and thesis defense, which would give them a better grasp on the subject matter.
What’s your advice to future data scientists who are considering a coding bootcamp?
At the end of the day, you cannot become a data scientist in 12 weeks, so you should learn the most relevant and important concepts. The most important thing is to keep learning after the bootcamp is over. NYC Data Science Academy has made me feel like I can maintain a lifelong commitment to learning.