In January to March of 2017, I attended Insight Data Science, a self-directed fellowship (not a bootcamp) which aims to help PhDs from many fields transition from academia to an industry career in data science. I emphasize that this is not a bootcamp; the organization recruits candidates that are well-versed in most, if not all, aspects of data science and as such, does not offer actual coursework. Therefore, participants are familiar with techniques such as: statistics, experimental design and A/B testing, machine learning, computer science and coding, and importantly, effectively communicating results to many different kinds of audiences. Therefore, Insight Data Science is more of a networking opportunity for driven individuals to get their foot in the door at prestigious companies looking to hire data scientists.
This 7-week program is divided into 3 phases:
1. Weeks one and two focus on developing a project. You can develop your own project, or work in collaboration with a company that is looking to make sense of their data. For example, I chose to work with an online financial company who was looking to predict click-conversion: in other words, what differentiated online users who chose to click and enter their contact information to obtain further product information, versus those who did not click and enter their information? This was a challenging project due to an extreme group imbalance; only 100 individuals from more than 2 million users chose to enter their contact information. I therefore had to account for this imbalance in my predictive models. (For more information about this project, click here.) The challenge here was to develop a project in significantly less time than we're used to as academics. We needed to be comfortable with creating a viable deliverable within tight deadlines.
2. Week three focuses on creating a 5-minute presentation of your project. This was also tricky; as academics, we are used to having a nice chunk of time to defend our work. It is quite difficult to condense the most important points of a complicated project within 5 minutes without using field-relevant jargon. This task allowed us to develop our communication skills for a general audience, a crucial skill for data scientists.
Throughout weeks one to three, fellows periodically listen to pitches from visiting companies who are currently hiring data scientists, develop their skill set with peer-led exercises in statistics, computer science and coding, and work on personal development, improving interview skills, resume, and online presence.
3. Weeks four to seven include visiting various companies to present your 5-minute presentation about your Insight project. We visited companies ranging from start-ups to well-established multi-billion dollar corporations in fields such as tech, media, finance, fashion, retail, healthcare, non-profits, consulting, and entertainment. If a company liked your presentation, they contacted you directly to schedule an interview, and hopefully, offer you a data science position with their company.
So what did I think of the program overall? This was probably one of the most demanding and intense undertakings I had ever accomplished. Having a background in experimental psychology, I felt that I was well prepared in some areas (e.g., statistics, communication) and woefully unprepared in others, such as computer science and coding. Therefore, I had to quickly immerse myself in these concepts and skills in order to complete my project within these tight deadlines, which was quite stressful. For psychologists who are thinking of applying to Insight, make sure that you are familiar with coding (preferably Python) and basic computer science concepts before applying!
Fortunately, I was lucky to be surrounded by exceptional Insight fellows and mentors who were all welcoming and supportive of each other. While learning is self-directed, it is also collaborative. You are encouraged to ask your cohort members with different areas of expertise for assistance. I thank all of my cohort fellows for being patient and instructing me on areas I clearly needed further development; they never made me feel unintelligent and simply conveyed the fact that we have all trained in different fields, and as such, have different strengths and weaknesses.
Participating in Insight was a wonderful experience, and I'm proud to call myself an Insight alumnus. While I was close to being offered a data science position, a number of extenuating circumstances arose which prohibited me from pursuing a data science career at the time. However, I remain passionate about data science, and still consider transitioning to the field. At the very least, I would like to apply data science concepts and skills to my current work.
Therefore, one of my goals for the coming year is to better understand and use Python. While familiar with R, Python is a true versatile programming language and is better able to handle engineering and production; it is also one of the most widely used programs in data science today. Psychologists traditionally use SPSS as their primary software for statistical analysis, likely due to its ease in usability. However, Python and R can conduct the exact same analyses and is more commonly used in industry. At the least, psychology students should be taught SPSS and an additional statistical programming language to better prepare them for careers in industry.
Also, I aim to familiarize myself with different machine learning models utilizing psychology-oriented data sets, with the hopes of demonstrating the applicability of data science techniques to psychologists. While machine learning models are slowly creeping into the field of psychiatry, they remain relatively absent in psychology (this recent Psychological Science article attempts to explain the utility of machine learning in psychological research.)
Periodically, I'll post tutorials on Python coding and machine learning models using psychology-related data. Hopefully, some of you may find this relevant and/or applicable to your own research. Stay tuned!