Statistical Data Analytics. Foundations for Data Mining, Informatics, and Knowledge Discovery
Christophe Lalanne · Journal of Statistical Software · 2016
This book is not a book on data mining, as the (sub)title may suggest.It sets out the principles of exploratory data analysis, statistical inference (Part 1) and basic building blocks for supervised and unsupervised learning (Part 2) with the R statistical package.In Chapter 2, the basic of probability theory and statistical distributions are discussed and illustred with R on real-world data sets.Chapters 3 and 4 are dedicated to data manipulation and data visualization, using R base graphics.The author emphasises the use of Tukey's fivenumber summary for descriptive statistics and outlier analysis, and of smoothing techniques for data with low signal-to-noise ratio.Most of the classical univariate and bivariate graphical displays are discussed at length, with relevant details about R's internals on, e.g., kernel density estimation or histogram binning.Chapter 5 deals with parameter estimation and classical test of hypotheses for the comparison of means and proportions.Interestingly, this chapter is not limited to maximum likelihood point and interval estimation.The author also briefly discusses the method of moments or the weighted least-squares approach, as well as simultaneous or bootstrap confidence intervals, among others.