Ethan Teo, an Economics student from Hughes Hall, discusses how you don’t have to be a CompSci to work in Data Science
Before coming to Cambridge, I was fortunate to receive opportunities that allowed me to explore various fields:
from collaborating with Singapore’s national defence research agency on a project when I was 16, to discovering more about the legal and investment banking sectors through attachment programmes with the Attorney General’s Chambers and UBS, to my enlistment in the Singapore Army at 18, and thereafter working for close to half a year at Singapore’s central bank and financial regulatory authority. Each experience –across defence science, law, finance, the military and the public sector – provided invaluable lessons and insights.
Last summer, I interned at the Institute of Data Science (IDS), a research institute at the National University of Singapore.
Some friends were surprised at my choice. Data science, after all, is mainly perceived to be dominated by computer science graduates and occasionally, engineers or statisticians. Indeed, these were the predominant fields my colleagues at IDS came from. However, while my background is in economics, I figured this could precisely provide a complementary perspective to the technical skill-sets of data scientists. I crafted my pitch and emailed a professor who I wished to work with at the Institute.
I learnt a great deal about the ways in which data science and economics reinforce each other
My two-month research attachment at IDS was rewarding and fun.
I learnt a great deal about the ways in which data science and economics reinforce each other – specifically the application of machine learning and artificial intelligence on behavioural economics. Take, for instance, the design of nudges in public policy which are aimed at inducing behavioural changes to enhance social welfare. While current interventions are mainly implemented in a one-size-fits-all manner, we can leverage on machine learning to develop a decision engine that is able to find insights into an individual’s unique combination of biases and heuristics based on their behavioural data. This then informs policy and allows us to support decision-making through personalised, tailored interventions – that is, targeting the right nudges for the right individual in the right context, on a dynamic self-optimising basis.
What I found most memorable about my stint at IDS were the interactions with my colleagues.
There was an informal, relaxed workplace culture that facilitated both independent and group learning. When a colleague knew I had recently completed an online course on machine learning and was looking for opportunities to apply my new knowledge, he brought me in to assist with his project. Using data from the Land Transport Authority, we worked on forecasting passenger traffic volumes for Singapore’s Mass Rapid Transit system. After pre-processing the raw data, we tried different models used in time series analysis such as Autoregressive Integrated Moving Average (ARIMA) and Support Vector Regression (SVR). While the initial results were not ideal, we worked on fine-tuning the machine learning procedure to further improve performance using cross validation and hyper parameter optimisation methods.
I would highly recommend exploring careers outside the obvious ones linked to your subject of study.
Besides being able to keep your options open, it allows you to make a more informed decision about which areas you might like to pursue in the long-term. In addition, there is a growing consensus that the future is interdisciplinary. From my internship at IDS, I believe the intersection of data science and economics/public policy is one such exciting development, and an area that I would definitely keep an eye out for.
A random fact about me…
I have two driving licenses – one for driving a car in Singapore, and the other for operating a military tank. It’s a pity neither are particularly useful for travelling in Cambridge to lectures!