Even though there are more flexible classification techniques, logistic regression remains popular. It's fast, it's interpretable, and it is much easier to do inference (such as constructing confidence intervals) other than prediction with logistic regression than more modern machine learning techniques. Although logistic regression is covered as a subtopic in other books, if you use it a lot you will benefit from a dedicated resource that gives application-specific advice.
Recommended Books
Applied Logistic Regression
David W. Hosmer Jr., Stanley Lemeshow, And Rodney X. Sturdivant
Key Features
- In-text exercises
Key Topics
- Area Under the Receiver Operator Characteristic (ROC) Curve
- Bayesian Logistic Regression
- Case-Control Studies
- Cohort Studies
- Confidence Intervals
- Diagnostics
- Fitting
- Goodness of Fit
- Hypothesis Testing
- Interaction
- Interpretation
- Logistic Regression for Correlated Data
- Matched Case-Control Studies
- Mediation
- Missing Data
- Multinomial Logistic Regression
- Multiple Logistic Regression
- Ordinal Logistic Regression
- Other Link Functions
- Propensity Score Methods
- Sample Size Issues
- Variable Selection
Description
This is an excellent practical guide for using logistic regression. As you would expect, construction and fitting of logistical regression are neatly introduced, as are the usual regression tests. More importantly, this book covers the interpretation of the model, including in the case of correlated data. Many useful fit diagnostics are discussed, and there is a useful discussion of alternative link functions and the Bayesian viewpoint on logistic regression (the Bayesian section could use some expansion). We found the exercises interesting, but there is little in the way of actual code support (there is some discussion of software packages). However, for most of the primary techniques, it isn't that hard to track down R packages that are suitable.
