Summary

In this first post of a three-part series, I reflect on my Master of Science in Systems Science degree. I explain why I chose systems science, and how that’s helped me become a data scientist.

I decided to become a data scientist two years ago. I had a bachelor’s degree in biology at the time but I didn’t want to get a PhD, so I was looking for something new. I’ve always been interested in cross-disciplinary topics (what drew me to biology was the complexity of the behavior of ant colonies; in retrospect, I was more interested in the complexity than the biology). Drew Conway’s venn diagram of data science inspired me: a discipline that was at the intersection of programming, math/statistics, and domain knowledge (and paid well, too!) sounded perfect. But I was a novice trying to break into the field, and as a novice I faced about six directions to go in:

• Master’s in Statistics

• Master’s in Computer Science

• Master’s in Systems Science

• Master’s in Data Science (online, e.g. U.C. Berkeley’s)

• Data science bootcamp

• Try to get a data analyst job and work my way up

I wanted an in-depth, in-person education (also, most job postings give favor to graduate degrees); these criteria ruled out the last three options. I spent a while deciding whether to pursue a master’s in statistics, computer science, or systems science. Statistics underpins machine learning and artificial intelligence, fields as hot as they get right now. Computer science teaches programming and software architecture, how machine learning algorithms get implemented and how code that works gets developed. Ultimately, I chose systems science because I wanted a broader field that combines computer science, math/statistics, and business.I recognized that I needed more concrete skills too, so I also enrolled in a graduate certificate in Business Intelligence and Analytics.

Systems science emerged in the 1940-1950s with cybernetics and information theory, chaos and complexity in the 1970s-1980s, artificial life and self-organized criticality in the 1990s-200s, and is now heavily focused on computational models of individuals, businesses, or networks for economics, psychology. One side of systems science is systems theory, which looks for isomorphisms of structure or behavior from different fields. This ‘stuff-free’ approach looks for patterns between systems that initially seem quite unalike.The classic example is that flocking birds and schooling fish behave with similar rules: the entire group moves cohesively by taking speed and direction cues from immediate neighbors. This is elegant, informative, and easy to model computationally (not so much analytically!), but ornithologists and ichthyologists don’t share overlap without systems thinking.

The other side of systems science is systems analysis, which I’m more drawn to. I’m a pragmatist at heart: I like finding patterns between problems, and identifying their solutions. Systems analysis deals with operations research, information theory, game theory, graph theory, and other disciplines in the pursuit of solutions to problems. Systems theory can show the similarity of different types of problem, but systems analysis goes further and tries to solve them. Elements of systems analysis show up in other departments including Statistics, Engineering & Technology Management, and Environmental Science & Management. Because the systems field is so broad, I could enroll in a variety of classes from these departments and count them towards my degree. I knew that I still needed to take courses in statistics and programming, and I happily did so. I also picked up another graduate certificate in Computer Modeling & Simulation to learn the hands-on skills for understanding complex systems.

Systems Science isn’t a typical path towards becoming a data scientist, but I think it’s a competitive advantage. In addition to learning data science skills (data visualization, univariate and multivariate data analysis, data warehousing, machine learning algorithms), I learned that cause and effect can be separated by both time and space, that phenomena we simplify to be analytically tractable can instead be understood more holistically with a computational model, and that for a decision to be effective, one needs to understand the environment and system one finds oneself in. By combining systems science and other disciplines, I created a curriculum that taught me the principles of data science while usefully expanding my view of how the world works.