Numerical Methods for Data Science Class

Although data science algorithms can be run by anyone who can push a button, to understand or innovate them requires a deeper understanding of the underlying mathematics. This special topics course will provide an introduction to the matrix numerical analysis techniques relevant to data science.

Topics to be covered include:

• Optimization via gradient descent, SGD, Newton-like, and alternating iteration methods.
• Matrix data and latent factor models: randomized approximations for SVD; non-negative matrix factorization; and matrix completion.
• Numerical methods for graph data: adjacency, Laplacian, and other graph matrices; spectral clustering and graph partitioning; centrality measures.

This course is being offered by the Mathematical Sciences Graduate Program at the College of Charleston but is suitable for advanced undergraduates as well as those who already have degrees in related fields. The pre-requisites for this graduate-level special topics course are only undergraduate courses in multivariable calculus and linear algebra. The course will meet on campus at CofC Tuesdays and Thursdays from 4:00 to 5:15PM.

For more information or for assistance registering, contact Alex Kasman ([email protected]).