Statistical modelling and machine learning offer a vast toolbox of inference methods with which to model the world, discover patterns and reach beyond the data to make predictions when the truth is not certain. This concise book provides a clear introduction to those tools and to the core ideas - probabilistic model, likelihood, prior, posterior, overfitting, underfitting, cross-validation - that unify them. Toy and real examples illustrate diverse applications ranging from biomedical data to treasure hunts, while the accompanying datasets and computational notebooks in R and Python encourage hands-on learning. Instructors can benefit from online lecture slides and solutions to all the exercises. Requiring only first-year university-level knowledge of calculus, probability and linear algebra, the book equips students in statistics, data science and machine learning, as well as those in quantitative applied and social science programmes, with the tools and conceptual foundations to explore more advanced techniques.
Statistical modelling and machine learning offer a vast toolbox of inference methods with which to model the world, discover patterns and reach beyond the data to make predictions when the truth is not certain. This concise book provides a clear introduction to those tools and to the core ideas - probabilistic model, likelihood, prior, posterior, overfitting, underfitting, cross-validation - that unify them. Toy and real examples illustrate diverse applications ranging from biomedical data to treasure hunts, while the accompanying datasets and computational notebooks in R and Python encourage hands-on learning. Instructors can benefit from online lecture slides and solutions to all the exercises. Requiring only first-year university-level knowledge of calculus, probability and linear algebra, the book equips students in statistics, data science and machine learning, as well as those in quantitative applied and social science programmes, with the tools and conceptual foundations to explore more advanced techniques.
Über den Autor
James Burridge is Professor of Probability and Statistical Physics at the University of Portsmouth, where he teaches probability, stochastic processes and statistical learning. He models language, birdsong, rocks, tessellations and games, and develops commercial applications of machine learning in green technology.
Inhaltsverzeichnis
1. Orientation; 2. Supervised learning warm-up; 3. Unsupervised learning warm-up; 4. Interlude: probability, likelihood and Bayes; 5. Probabilistic modelling; 6. Frequentist and Bayesian uncertainty; 7. Frequentist linear regression; 8. Directed graphical models; 9. Bayesian linear regression, priors, and regularisation; 10. Bayesian methods; 11. Classification; 12. Unsupervised learning: a deeper dive; 13. Neural networks and deep learning; 14. Expanding the toolkit; A. Probability theory; B. Linear algebra; C. Jensen's and Gibbs' inequalities; References; Index.
