You are here

The Manga Guide to Regression Analysis

Shin Takahashi
No Starch Press
Publication Date: 
Number of Pages: 
The Manga Guide Series
[Reviewed by
Adam Gilbert
, on

Allow me to begin by saying that I grew to really enjoy The Manga Guide to Regression Analysis. The approach to explaining regression analysis through the manga storyboard was quite unique and effective. The story follows Risa and Miu, two employees at NORN’s Tea Room, and discusses the mathematics through a series of tutoring sessions. Miu is a very intelligent but unconfident girl who wants to learn regression analysis so that she can speak with a boy who accidentally left his regression textbook at NORN’s one evening. While I appreciate the fact that Risa is an extremely intelligent and confident girl, I wish that the book’s story line was something less stereotypical than “girl learns math to impress cute boy”. I am looking forward to future books in this series which will hopefully break such gender stereotypes.

The Manga Guide to Regression Analysis begins with a review of the mathematics necessary for discussing regression, followed by chapters on simple linear regression, multiple regression, and logistic regression. All of the major problems I had with the book were contained within the first chapter, “A Refreshing Glass of Math.” A few things bothered me here, such as Risa stating that the value of e is 2.7182 (p. 19) and also the presence of incorrectly stated rules for differentiating the common logarithm (p. 36). I suspect that these rules had been intended to be written for the natural logarithm since in all other instances the book references the logarithm base e. Despite these shortcomings, there were some nice items contained within this chapter. The short review of statistics was particularly well done, including measures of central tendency and variation, hypothesis testing, probability density functions, and distribution tables. The reader in need of a more thorough review of introductory statistics is referred to The Manga Guide to Statistics.

Once we reach the chapters on regression analysis, the book becomes quite impressive. Throughout the chapters, Risa and Miu explore regression as a way to help NORNS (and in one case, a similar business) make well-informed decisions. These applications of regression include the number of iced teas NORNS is expected to sell (simple regression), how much revenue a chain can expect a potential new storefront to generate (multiple regression), and also whether or not a particularly expensive perishable product will sell on a given day (logistic regression). The use of such tangible examples will be quite helpful to the reader. These examples also readily showcase the utility of regression analysis.

Each chapter begins with an overview of the problem at hand, a very “high-level” discussion of the proposed regression technique, a clear outline of the necessary steps involved, and then the complete regression procedure is worked out. I really like that the author includes the procedural outlines and found them quite useful to refer back to. The outline also helps to keep the regression process from becoming overwhelming. Students will benefit greatly from this. I also liked that this book does not shy away from calculations, particularly in the simple regression chapter. The calculations are explained quite well, are very well organized (typically in tabular format), and are easily followed.

There were a few instances where I thought the book could be improved. First and foremost, I wish that probability distribution tables were included as an appendix. This would make it much easier for the reader to verify the conclusions to each application. The book missed opportunities to discuss working with categorical predictors that have more than just two levels. For the most part, however, the book does quite an admirable job. Upon finishing it I couldn’t believe how much of regression analysis was actually covered within this approximately 200 page story. The tutorial on regression in Excel, included as an appendix at the end of the book, will be of use to the general reader.

The short of it is, this book is good, but far from perfect. A disappointing storyline, some mathematical missteps, and missed opportunities are its shortcomings. That being said, the book does an admirable job discussing regression techniques in a way that is engaging, accessible, and informative. This book is certainly not a textbook, but for the right reader, it could make a great companion to one.

Adam Gilbert is an Assistant Professor of Mathematics at Southern New Hampshire University in Manchester, NH. His research interests are in graph theory and combinatorial games. He enjoys using regression and data analytics techniques in applied research projects with undergraduate students.

The table of contents is not available.