When I heard about the work at UNDP Global Pulse, I thought it was something straight out of a movie. Data science for social impact sounds like the coolesting thing I can do with mathematics.
I developed my passion for mathematics, economics and computer science at different stages of my life. At UCLA, I have to put on different hats throughout the day as I switch between writing mathematical proofs, studying economic theories, and writing computer programs. I've always pursued these three exciting fields separately until this summer.
Here are some of the lessons I learned about research, data, the economy and myself during my time at UN Global Pulse Lab Jakarta (PLJ).
1. You can reduce almost every problem to a mathematical problem
I had to implement clustering algorithms within a week on hundreds of millions of datapoints with zero knowledge in machine learning. Fortunately, I'm a math majorâ??-â??My math background really helps me to build my understanding of the algorithms used from the ground up.
For instance, understanding the linear algebra behind Principal Components Analysis (PCA) pays off when I interpret the pretty visualizations from R to make decisions on feature selection. The idea of PCA makes sense to me when it is described "as a mechanism to recognize geometric similarity through algebraic means" (Jauregui,2012) , or, the way I perceived it, as mapping data points to the most significant 'direction' as determined by the eigenvectors of covariance (correlation?) matrix of the dataset's independent variables. The concepts make sense to me because the math behind them does.
I gained abundance of appreciation to mathematics because I've applied them at my fingertips to real-life projects. It's like my 'why' moments last quarter (Why do I need to orthonormalize eigenvectors? Why are there a million theorems on eigenvalues? etc.) were answered and these were priceless revelations.
But real-life problems are more than just math problems. There are many other layers of moving parts that need to be considered.
2. Intersect the (seemingly) opposites
I've never met a team quite as diverse as PLJ. I worked in the quant research team, however I often found myself sitting in social systems team's workshops and lectures about urban planning, marine biology and socio-economic surveys amongst many other topics.
PLJ's Research Dive exemplified how deeply intertwined the two fields of research in data science for public sector. The hackathon-styled event, titled 'Machine Learning to Estimate Poverty', gathered esteemed researches across the nation for 3-day data exploration and analysis to learn more about poverty in Indonesia. It was thought-provoking to watch statisticians and computer scientists work on the same research project as social scientists and government officials, and witness their expertise to build on one another.
Engaging experts from seemingly unrelated fields proves to be an insightful practice for me. Upon presenting my project to the lab, I was guided by advices from the social systems program managers to qualitatively define the objective and impacts from my analysis. Their advice directed me to refine my research objective and consider the implications of my findings as might be perceived by policy-makers.
This was one of my favorite experiences during my internship, and maybe even the field in general: connecting the unlikely dots and assigning some probability to them.
3. Open Data, Open Innovation, Open Government...
We live in a world where sharing is caring. Sharing knowledge creates synergy and market efficiency. Innovation and data matter most when we get them to the right hands.
I was introduced to the concept of open government through the works of other members of the quant research team, who have worked for Open Government Indonesia and are conducting measures of data openness in Indonesian cities. I was amazed by the progress made by Indonesia as an emerging economy to become a more open economy. Organizations like Satu Data Indonesia, BappenasRI, Jakarta Smart City, and many more are close collaborators with the lab, and their work have also inspired me immensely about 'openness' at an institutional level.
I learned that this openness needs to be fostered in the form of collaboration and bridging dialogues between various stakeholders, both internal and external. There is much work to be done as sharing knowledge (especially conversations around what to share, how to share, why we should shareâ?¦) tends to be underrated. To implement tools of big data efficiently and productively, a robust big data infrastructure is needed, and ideally should be considered as a public good.
4. Handshakes and Hand-Offs
That is, " the handshakes between powerful, enabling entities (read: public sector) that allow for the handoffs to those with the hope, ambition, inspiration, and ideas (read: private sector) to make our country better". I encountered this idea while reading Aneesh Chopra's 'Innovative State', which has earned its place as one of my all-time favorite books. The ideas discussed in the book have helped me understand the potential of technology in public sector, and also the role of government as a tool to further the technology landscape. Having an open government, as mentioned above, is one way.
I started to see what the quote actually means at the lab. Although UN Global Pulse is not a private company, the organization adopts the relevant technology and the curious mindset similar to most start-ups. However, the team has a responsibility of advocating for technology and integrating them with existing governmental/societal structures. PLJ closely works with the government to build data-driven platforms for policy-making. The task demands a diverse skill set to consistently juggle various incentives and delegate tasks of different subject fields, all built on mutual trust. No easy feat. This requires commitment and respect, atop of a high level of technical expertise. It is both technically and socially challenging.
5. Out-work your fear and insecurities (impostor syndrome is too real!)
I was the youngest, least experienced and least skilled person in the lab, which initially made me intimidated and insecure. At the same time, I felt really lucky and humbled to learn from such brilliant and passionate peers everyday, and I choose to embrace this feeling.
I turned to productivity and optimism to diminish my impostor syndrome. I worked tirelessly to listen, to ask questions, to study ahead and to persevere until I get the appropriate answers and solve the problems. Googling things, not being afraid to ask and never giving up got myself to the right places.
I personally found that one of the best habits for continual improvement is to put yourself out of your comfort zone. The key is to stay strong and centered during this process.
This post is only a brief recall of my experience and these lessons are just starting points of my journey in data science. I've always loved mathematics and creative problem-solving, but I learned that I desire to uphold a purpose in my career to contribute to social good and/or economic development. When my personal values are aligned with the team's mission, I feel good going to work every morning.