This is the third in a weekly series spotlighting different members of the 66DaysofData discord community, an initiative started by data scientist and YouTuber Ken Jee. You can find out more about the initiative here and here. The intent of this series is to celebrate those who have made it through the challenge already (although it never truly ends!); share their stories; learn from their successes and failures; and hopefully inspire anybody else walking this path, wherever you happen to be in your journey. If you’re interested in joining the community, you can do so here.
The bad news is data science is incredibly broad and not as well defined as it maybe should be.
The good news is data science is incredibly broad and not as well defined as it maybe should be.
Still an electrical and computer engineering student, Brian Doan was introduced to data science when he came across Ken Jee’s channel in June of 2020. He had just started to get a taste of electrical engineering through an internship at the time and discovered that for him personally, some aspects of practical implementation were not necessarily as fascinating as the theoretical underpinnings. On the other hand, data science was an exciting unknown, with so much to explore and tinker with that he found it hard to get bored. Brian’s educational background came in handy when he decided to join the 66 days of data challenge in September.
Brian started with Kaggle’s mini courses, then started looking for other resources, digging into the Data Science Handbook before moving onto the super popular and often recommended Hands on Machine Learning. On top of being a full-time student, he also found the time for multiple fun side projects during the first round of the 66 days of data. These projects ranged from a trading strategy model to YouTube sentiment analysis to Pokemon team building; some even used datasets he’d curated himself. Hidden in all of this is Brian’s first secret to his success:
Working on his projects in parallel as opposed to in series helped him keep his momentum and motivation up.
This was particularly illuminating to me because I’d subconsciously started thinking of project’s status as binary: either it’s complete or it’s incomplete. Admittedly, a not-insignificant part of my excitement about finishing my first data project was a fixation on the finish line. As I crawled along toward the “end”, I started feeling more of a self-imposed pressure, which took away some of the enjoyment as my focus shifted away from the process and toward the outcome.
Committing to a deadline publicly also works for some people to boost their motivation. If you want to supercharge it though, Brian found a creative way to up the ante: List a project you’re currently working on and committed to on your resume as a work in progress! However, use this one selectively and with discretion; make sure it’s a project with some substance that you can talk intelligently about already. This does a few things:
· You now have one more person holding you accountable to finish your project.
· You’ve created artificial deadlines for completion.
· If you complete it by the time you get an interview, that shows progress. If you haven’t, it’ll still be a fresh topic that may lead to an interesting conversation with your interviewer.
Speaking of the job search, this brings me to the next gem from him:
Collecting unique or different data is nice, but creating unique or different insights is even more important.
To be clear, we’re not talking about making Nobel-prize winning discoveries; we’re talking about applying your own diverse set of knowledge and experiences to put your own spin on your data projects. While he’d created his own datasets for a couple projects, Brian was able to land internships during the first round of the 66 days of data by showcasing projects built using Kaggle datasets (no, not the Titanic or Boston housing datasets). Focus more on the “so what?” and the impact of your work.
Currently, in the second round of the challenge, he’s decided to take the next step and work on model deployment, through the ever-popular Streamlit, an open-source app framework. True to form, he has no qualms about going back to Kaggle for his dataset. Who knows, we may even see him writing articles on his progress this time on Medium, after an exchange he had with another article writer. After reaching out to the writer because their article resonated with him, the writer asked Brian whether he’d considered writing his own. Great question, I think.