Statistical Learning for Big Dependent Data
María Dolores Ugarte · Journal of the Royal Statistical Society Series A (Statistics in Society) · 2022
This is an interesting and well-written book on learning from univariate and multivariate time series data mainly, although the book has a final chapter on spatiotemporal data as well. This book has been written by two experts in the field of time series (they both have written books and many scientific papers on this topic) and, as the own authors recognize, it does not pretend to cover all the aspects related to big dependent data. The authors have chosen certain material mostly because of personal experience and preferences. For example, topics related to the analysis of functional dependent data or data taken at irregular time points are not covered in this book. What the authors understand by big data is also informative for potential readers. In this book, big data means that the number of analysed variables is large or the number of data points is also large or both, and it could happen that the number of variables is greater than the number of data points. The title of the book could well have been Statistical Learning for Big Time Series Data rather than Statistical Learning from Big Dependent Data as the first eight chapters of a total of nine are related to the analysis, classification, and forecasting of time dependent data. These chapters are very well organized and they are concise and highly informative. They include R scripts to reproduce examples and this is appreciated. The R package SLDBB accompanies the book and can be found in CRAN (https://cran.r-project.org/). Some data sets are also provided at https://www.rueytsay.com/slbddb. Only the final chapter deals with spatio-temporal data. This chapter is less comprehensive and describes mainly kriging methods. I enjoy reading this text and I think it is a nice addition to the scarce literature on analysing big time series data. However, although the authors try to start from the very beginning in the analysis of time series data, I would not recommend this book for beginners but for readers with at least some prior knowledge on stochastic processes and time series analysis as well as a good basis in matrix algebra.