A Programmer's Guide to Data Mining

Livro digital

Título:
A Programmer's Guide to Data Mining

Autor:
Ron Zacharski

Categoria:
Tecnologia > Dados

Doador:
Raffaello D. N.

Sinopse:
Recommendation systems become far less mysterious when their machinery is built from first principles. Ron Zacharski starts with user-based collaborative filtering, sharpens it through implicit ratings and item-based filtering, and then progresses to content-based classification, k-nearest neighbors, Naive Bayes, text classification, and clustering. The sequence turns the familiar experience of receiving a recommendation into a practical path through core data-mining ideas. Concrete examples connect similarity measures, normalization, evaluation, probability, unstructured text, and group discovery to problems readers already recognize from music services, online stores, social data, and medical prediction. Executable code and small datasets keep the mathematics close to observable behavior, while discussions of sparsity, popularity bias, scalability, and algorithm choice expose why a technique that looks convincing on paper may fail in a real system. The result is an unusually approachable bridge between programming and machine learning. Readers do not need advanced mathematics to begin experimenting, but they leave with enough conceptual discipline to compare models, question results, and build recommenders or classifiers of their own. It is especially useful for developers who learn best by implementing an idea, testing it against data, and understanding where its assumptions break.