Customer Reviews

Fitting Models to Biological Data Using Linear and Nonlinear Regression: A Practical Guide to Curve Fitting

5 star:		(3)
4 star:		(0)
3 star:		(0)
2 star:		(0)
1 star:		(0)

 

 

I've been looking for a book like this for 5 years to help me understand better nonlinear regression. The only choices before this were the same old basic stats books, or the other extreme of stats books for mathematicians. As an engineer, I needed a book that does APPLIED, not THEORETICAL, nonlinear regression. This book gives examples and speaks normal English, unlike Seber and Wild's book, which is virtually devoid of examples and drowns one with matrix math instead. Seber and Wild's book is more like a dictionary of stats equations. For Motulsky's book, here are 3 examples of things you won't find in most other statistics books: 1) the difference between confidence bands and prediction bands for Y(p. 32), showing that the former doesn't include a majority of the points, while the latter does; 2) How to compute egg- and elliptical-shaped joint confidence regions for 2 parameters (p. 114-121); 3) Good explanation of 3 different ways to compute parameter conf. regions (asymptotic, Monte Carlo, model comparison). Just one of these 3 makes it worth the price of the book for me.
Motulsky has so many different helpful topics in here, that he obviously has run into many of the problems that nonlinear regression people will see. The book is helpful regardless of the software you use, because I do a lot of coding in Matlab. I may make this book a standard text for a new engineering statistics class I am preparing.
The main shortcoming of the book for me (remember that I'm an engineer) is lack of the exact equations 1) for confidence bands and prediction bands for Y (p.32), 2) for asymptotic conf. interval for parameter (p. 98), and 3) for standard error vs. standard deviation. For the first 2, the author could reference eqn. numbers in Seber and Wild or Bates so his text flow would not be interrupted. For those of us who publish, we need to know what's going on in the black box. The standard error/standard dev. formulas would make the explanation given easier to follow.
In summary, this is the book that will help walk you through many of the problems and concerns in nonlinear regression. I'm glad someone who understands my situation finally wrote it!

The translation of experimental data into testable hypotheses rests, in large part, on the ability to describe the data in quantitative terms, and to model the relationship between experimental variables and outcome. In this exceptionally readable text, the authors provide a very practical approach to curve fitting routines for the graphic display of experimental data. From this basis, the authors guide the reader through the process of model building, model comparisons and model testing. All of this is a vital part of modern science, from basic biology to physics to chemistry and beyond to applied sciences such as drug discovery. The illustrative examples used throughout the text tend to derive from the authors' interest in pharmacology, but they are nevertheless useful and clear to anyone with a basis understanding of science and mathematics. This book should be part of the library of every serious scientist and student of experimental science. It offers an approachable introduction to modern data analysis.

I work in academia, in a very close collaboration with pharmaceutical industry, my interest is enzymology and drug design, I am a PhD scientist. The book is well written, easy to follow for a novice and an expert, all concepts are well illustrated and explained. Simply if you can not follow the material in this book you are not likely to follow anything. I am grateful to the authors for writing this book.