
06193877
b
2015a.00802
Levine, Evan S.
Applying analytics. A practical introduction.
Boca Raton, FL: CRC Press (ISBN 9781466557185/hbk; 9781466557192/ebook). xvii, 272~p. (2014).
2014
Boca Raton, FL: CRC Press
EN
K45
K75
K85
analytics
handbook
doi:10.1201/b15059
With this textbook, E. S. Levine offers a supplement to students who are beginning graduate work in quantitative data analysis. The intent is to outline important techniques, indicate appropriate and inappropriate contexts and interpretations, and help students who may feel themselves overwhelmed when dropped into a situation with an overwhelming amount of data and no experience interpreting it. Thus, it begins at an extremely basic level, not only providing formulas for basic statistical concepts such as the mean, but even explaining such basic information as to what constitutes analytics, how to decide which questions to investigate, and how to communicate one's results. According to Levine, vocabulary in the field has not been standardized, and little background on useful practice is available; most of the learning is done via experience; see, for example, the third paragraph of the Preface, but this recurs throughout the text, and Levine points out that ``uncertainty'' in definitions of important terms can lead to ``systematic error'' in communicating findings and conclusions. Mathematicians and computer scientists interested in analytics will find many familiar terms in this text, but quite a few terms are used in unfamiliar ways. In Levine's vocabulary, for instance, order in a ``list'' of data doesn't matter, which sounds an awful lot like a mathematician's or computer scientist's notion of a ``set'', while a computer scientist's notion of a ``list'', wherein order matters, is here called a ``1dimensional data set''. The author gives the reader no hint of this curious reversal in the text, but the reader adapts fairly easily. As indicated above, the text begins with very basic exposition, but covers some fairly sophisticated topics, such as Fourier analysis, wavelets, text mining, and Bayesian analysis. Levine eschews justification for the formulas, as this lies beyond the scope of a supplementary text. In any case, frequent citation of other works, and encouragement to visit them for further information, should help the reader interested in deeper ideas. The text does contain wellconsidered examples that go a long way towards explaining many of the ideas. Only one explicit data set is given the reader, and that one artificial; Levine uses it repeatedly throughout the text, and to very good effect, in this reviewer's opinion precisely because of its artificiality, as Levine can use it to illustrate the importance of thinking about outliers and probability distributions. Other examples are pulled from publiclyavailable datasets, such as the U.S. Census, though they are not explicitly given, as they would be too large. Levine also comments insightfully on realworld situations, such as stock market prices and the 20072008 collapse of the credit market, beginning in the housing sector.
John Perry (Hattiesburg)