In 2019, he published this textbook to bridge the gap between traditional math and these new applications. The book serves as the foundation for his MIT course 18.065 , focusing on how data is reduced and interpreted through matrix methods. Key Themes of the Book
Strang also introduces the concept of matrix norms (Frobenius, spectral, nuclear) as objective functions for learning—a topic absent from classical texts.
After finishing Part III, go back and re-express everything you learned (PCA, least squares, regularization) as variations of the SVD. Strang argues that if you deeply understand the SVD, you understand 80% of data science.
Do not skip this section, even if you think you know linear algebra. Strang reviews the fundamentals through the lens of data. Key topics include:
Do not just do exercises. Take a dataset (e.g., MNIST digits, housing prices, movie ratings) and apply the methods:
