The subdivisions or "Chapters" are clusters of knowledge around Numerical Mathematics, more or less strictly ordered by increasing specificity or involvedness. Use this to guide your learning journey around anything that involves calculations on computers with limited precision (the vast majority of cases). This is relevant for instance for complex computations involving a very large number of multiplication or division operations, especially when the sought values are in the [0..1] range.
Practical relevance for this lies in the training of neural networks and their deployment on less capable infrastructure, for instance. Other applications revolve around eigenvalue, singular value computations etc. as part of numerical simulations aka. interpolation, approximation, FEMs and other methods.