Foundations of Data Science

Livro digital

Título:
Foundations of Data Science

Autor:
Avrim Blum, John Hopcroft, Ravindran Kannan

Categoria:
Tecnologia > Dados

Doador:
Raffaello D. N.

Sinopse:
This book opens with the geometry of high-dimensional space, then moves straight into best-fit subspaces, singular value decomposition, and the power method, so it meets the reader where data science becomes mathematically real: when intuition has to survive large dimensions and noisy data. From there, the table of contents lays out a deliberately broad toolkit, including random walks and Markov chains, Monte Carlo methods, principal component analysis, clustering mixtures of spherical Gaussians, ranking documents and web pages, and a full machine-learning arc that reaches perceptrons, kernel functions, regularization, online learning, SVMs, and VC-dimension. The later chapters extend that foundation into streaming, sketching, sampling, clustering, random graphs, and topic models, showing how the same core ideas reappear across modern data problems. The result is a rigorous, connected treatment of the field rather than a loose survey. Readers get both the algorithms and the theory behind them, along with enough structural depth to understand why the methods work and where they are useful, making this especially valuable for anyone who wants data science as a serious mathematical discipline.