With a PhD in Mathematics, Pokman Cheung was no stranger to quantitative analysis, but he wanted to transition into a new career as a Data Scientist, so he enrolled in NYC Data Science Academy to get a grasp on the practical aspects of data science and machine learning. We sat down with Pokman to learn about his experience at the data science bootcamp, the diverse backgrounds of his classmates, and how he landed his new job at Goldman Sachs London!
Pokman also contributes to the NYC Data Science Academy blog- check out his post on Facial Image Analysis.
What were you up to before you started at NYC Data Science?
I had obtained a PhD in mathematics from Stanford, and held academic positions at MIT and Sheffield. However, I decided recently to pursue a new career direction in or related to data science.
Did you have a technical background before you applied? Had you taken a CS/math class, tried Codecademy or another online platform?
I didn't have any relevant technical background from previous work experience. Before applying for the Bootcamp, I had taken several online courses from Coursera and edX on data science and programming. While they provided good overviews of the subjects, I realized that I needed to find another way to gain a deeper understanding and some practical experience.
What was your goal in doing the NYC Data Science Academy bootcamp?
My goal was to get a deeper understanding and some practical experience in data science and machine learning, in order to be able to find a desirable data scientist job.
Why did you choose NYC Data Science? What factors did you consider? Did you look at other bootcamps or only NYC Data Science?
I have looked into similar courses. NYC Data Science Academy appealed to me the most mainly because of their comprehensive and practical curriculum, and their strong industry connections.
What was the NYC Data Science application like for you?
The application consisted of some coding problems and a phone interview. It helped me confirm that I am a good fit for the course, and understand what I am expected for and I can expect from the course.
How many people were in your cohort? Did you think it was a diverse cohort in terms of age, gender, and race?
There were 18 students. This is perhaps the most diverse group of people I have ever studied or worked with, certainly in terms of age, gender and race, but also especially in terms of background and experience. While everyone possessed at least the required technical level, the diversity in background and experience enabled many meaningful and fruitful interactions between the students.
Who were your instructors? What was the teaching style like and how did it work with your learning style?
Vivian, the founder of the Data Science Academy, has vast knowledge of the data science industry and a highly practical perspective. Her ability to share such knowledge and perspective, in the form of class lectures and detailed individual feedback, was in my opinion her greatest value. The other instructors came from such background as healthcare industry, Google and academia. They all share Vivian's practical and interactive style, with particular strengths in various aspects of data science.
What technologies did you learn in your course? Were you able to learn it all in the short time you were in your program?
The course covered various tools and techniques in data extraction (including web scraping), data cleaning, visualization and machine learning. These tools and techniques are mostly implemented in the languages of Python and R. It was a large amount of material, but the instructors made sure that we were able to absorb all of it through well-designed homework assignments, projects and discussion sessions.
Were there exams/assessments? What happened if you failed one?
There were daily homework assignments and four projects throughout the bootcamp, but no exams. We were given extensive feedback on our work.
Are there things you didn’t expect or that you would change?
The bootcamp met, and in some aspects exceeded, my expectations. In retrospect, I would like to have done a little more preparation beforehand in order to even more fully take advantage of the entire experience. The teaching and administrative staff constantly encouraged and responded to students' feedback. In particular, a student-staff meeting -- nicknamed `therapy session' -- was held every Thursday, and any useful ideas and suggestions brought up in the meeting would often be incorporated starting as soon as the following week.
Can you tell us about a project you worked on? What type of data set did you work with, which technologies did you use, what did you find out/discover, did you work on it alone or with a group, is it live now?
After having learned a fair amount of machine learning, the students were divided into teams of four or five and each team started working on a Kaggle competition of their choice. My team chose a competition posed by Ponpare -- Japan's answer to Groupon -- whose goal was to predict the coupons purchased by each user within a one-week period. The provided data included details of the users, details of the coupons, and all the transactions within the previous 51-week period.
Our initial attempt was to train a coupon classifier for each user using some classification method (e.g. neural network, support vector machine). However, poor properties of the resulting models led to the realization that our approach was inadequate in such a situation, where no user would purhase any more than a tiny fraction of all the available coupons.
The approach we eventually adopted was based on quantifying how similar two coupons are using cosine similarity. To achieve an optimal model, we utilized such techniques as feature transformation and cross validation. Our highest score once ranked 7th on the leaderboard.
What was the most challenging part of the course?
I found the projects to be the most challenging but also the most important part of the course, because they required us to not only utilize everything we had learned, but also find or choose our own methods.
Did NYC Data Science do job prep with your class- interview practice, resume building etc?
The NYC Data Science did a great deal to help the students find jobs. Throughout the bootcamp, they invited many industry experts to give talks and provide advice on job application. In the meantime, they also gathered and organized our coursework into personal profiles, which they used to promote us as candidates for suitable openings. Towards the end of the course, there were even more career-oriented activities like 'elevator pitches', meetups with recruiters, interview practice, and resume consultation.
What are you doing now- did you move up in your career or get a new job?
I will start a new job in September, working in the risk management department at Goldman Sachs London. This will be my first job in finance, with a significantly higher salary than my previous jobs in academia.
I started applying for jobs some time before starting the bootcamp, but only received an offer in the middle of it. The bootcamp was useful for my interview preparation, and also gave a positive impression of me to employers.
How long did it take to get the job at Goldman Sachs?
My job application started in February and last until July, a month after the bootcamp started.
Did you feel prepared for the interview with your current company? What is a Data Science role interview like?
I felt well-prepared for the interview. The final interview lasted for a whole day. It consisted of a presentation of a past project and meetings with five people. Besides the typical motivational questions, I was also asked a variety of technical questions, covering such topics as statistical inference, regression models, algorithms and codes, as well as some actual problems arising in my interviewers' work.
Was NYC Data Science worth the money? Would you recommend it? Could you learn that on your own?
I think so. The main values of the bootcamp are: (i) the instructors' knowledge and perspective in the industry, e.g. concerning which ones of the vast number of available tools and techniques are more important than others, (ii) the opportunity to interact with many established data scientists, and (iii) the experience of working on real-world data science projects with guidance from the instructors and collaboration with fellow aspiring data scientists.