Critical Analysis of Scientific Data

3 units

Data science is most commonly associated with topics in computer science. However, efficient algorithms, specific software packages, neural nets, and so on are only tools, and they are easily misused. In a research setting, working with data is primarily an exercise in critical thinking. The purpose of this interactive, hands-on course is to learn from mistakes by making them in a safe environment. After covering/reviewing probability theory, Bayesian inference, elements of information theory, and random matrix theory, the course will focus on case studies of real-world biological data, such as quantitative imaging data, next generation sequencing (metagenomics), and neural recordings. These modules will involve the critical reading of research papers and working through puzzle-based assignments. The primary modules will be supplemented by shorter presentations on topics chosen by students. Fair warning: This is explicitly not a course on big data or machine learning, although students may choose to explore some of these topics in their presentations (required for credit). Experience with MATLAB or Python strongly encouraged or will need to be acquired during the course. Open to undergraduates with prior programming experience and a quantitative background. Experience with data or statistics not required. Course mimics a research environment, and undergraduates considering an academic research track are especially encouraged.