Machine Learning, Data Science, and Algorithms Subcommittee

For the purpose of the Common Ground, the Steering Committee considers subjects in the context of the following definition and examples.

Machine learning and data science are about designing algorithms that can automatically extract structure from data and use data to find better solutions to tasks. Many disciplines are undergoing important shifts whereby perspectives from machine learning and data science can be used to rethink how to solve some of the basic problems. Accordingly, classes in this area will have a range of styles and focuses. A course should generally cover multiple aspects of the pipeline from data to decisions, such as:

Exploratory data analyses to understand data’s limitations and biases.
Translating the problem into a form amenable to computation. Examples include choosing an objective, positing a generative model, or making causal assumptions.
Methods for finding a solution. One example would be designing and implementing algorithms to fit a model to data.
Evaluating the performance, methods for understanding and interpreting the solution, and communicating the results to decision makers. An example of an existing course in this area is Modeling with Machine Learning: From Algorithms to Applications (6.C01) and its accompanying disciplinary modules, such as Physical Systems Modeling and Design Using Machine Learning (1.C01). Examples of desired new common ground courses in this area include undergraduate-level causal inference, optimization, data visualization with communication, and computing with big data.