Regression Modeling Strategies: With Applications to Linear Models, Logistic and Ordinal Regression, and Survival Analysis (Springer Series in Statistics)

I bought this text after using and learning about Professor Harrell's contributions through the literature and through the R and S computing communities. Others have written how wonderful Professor Harrell's software is to use, and how carefully he has thought about the entire cycle of doing statistics, including transparent analyses and publication using LaTeX.But the book is a joy, balancing theoretical concerns, a masterful selection of literature, and solid assessments and presentations of actual examples. Professor Harrell demonstrates how powerful these techniques are, and the material moving them. These insights are key in a time when increasingly people want to just dump numbers into a package and get something out of it, an analytical behavior often justified by cost and time pressures. (If it's wrong results you want, it's trivially easy to get those, even if they LOOK wrong.) Professor Harrell shows it needn't be that hard, with tools from the CRAN (the R community source), many good ones which he and colleagues contributed. But he also shows that it's important to keep an eye on the parts of these, and be wary of pitfalls.Professor Harrell is also candid in his assessments, even after giving enthusiasts for a technique he critiques the benefit of the doubt. I find his comparison of cross-validation with bootstrap validation wonderful, and his discussion of standard assessments of models like R^2 refreshing.Check out his lectures, too:[...][...][...]

My initial temptation is to say this is the best statistics text ever, but it's all relative. It perfectly suits my current needs and state of development. The book claims to be intended for graduate level students in biostatistics and I think that is a fair assessment (I am self-taught, so how am I to know?).I haven't even finished yet, but I am reading the text cover-to-cover after first perusing parts of chapter 10. This linear approach is facilitated by Prof. Harrell's excellent writing style.The text has a practical bent, but with plenty of theory and references to back up the practical advice. You may find Harrell's views to be controversial. I have been forced to reconsider many of my notions about model-building.I note that "r programming language" is a suggested tag for this product. While Harrell's Design and Hmisc packages are available to R users, the text actually refers to the use of S-PLUS and there may be subtle distinctions. As a Stata user, they're both alien to me, but this hasn't affected my enjoyment of the book.

I found "Regression Modeling Strategies" to be a fantastic treatment of a wide assortment of model selection techniques. Harrell's writing style is quite lucid (assuming you've had graduate-level statistics coursework). Model selection/validation is arguably the most critical component of the statistical literature for many industry statisticians, and it is rare to find a textbook solely devoted to the merging of theory with practice. This is not to discredit other applied statistical texts; they represent a necessary foundation to master before a text like Harrell's can be understood with any depth.It is often said that "All models are wrong but some are useful". To that I would follow with, "In the land of the blind, the one-eyed man is king". Harrell's text will help empower you as a statistical modeler. Personally, I think combining this book with Gelman and Hill's "Data Analysis" text creates about as good of a 1-2 punch that an applied statistician will ever find.

I've only read one chapter in this book, but I think I can say that this is one of my favorite math books. My experience with it has almost been interactive because all the questions that I have are always answered either in the next sentence or in the footnotes. Regression Modeling Strategies is definitely a steal for whatever they are asking, if not for the amazing content and references, but for the number of pages and colored pictures alone. Writing this review is sort of funny because even if I was lying, it doesn't really matter since if you are reading this, you probably need it. The biggest downside to the book is the subpar cover; boring mustard color with text.

It's been about two months since I bought the book. I have been reading it whenever I have time to. Till now, it's one of the best books I have read that addresses a difficult subject like regression model building.

The book is fairly complicated. Buyer beware

As logistic regression is the most used statistical method in health research, this book is very very useful because all of the case studies are about health research. Very useful book for health researcher...

I'm a graduate student in the life sciences and was looking for a book on multiple logistic regressions. My advisor suggest this book and I have not been disappointed. Harrell does a good job of balancing theory with application throughout the book. The inclusion of S-Plus/R code was also beneficial. Furthermore, I was impressed that the code still worked despite the booking being over a decade old!

Frank Harrell's way of explaining methods along with R code is very useful, even for beginners.

Eu fiz uma graduação em estatística e quando eu fiz a disciplina de regressão, ou mesmo as disciplinas de projetos finais, ter lido esse livro teria feito toda a diferença. Esse livro aborda diversos fundamentos da modelagem que não são ensinados tradicionalmente em um curso de regressão. Alguns conhecimentos que impactaram bastante minha forma de trabalhar com modelagem foram:1 - como a discretização tem um impacto negativo na modelagem;2 - a importância de utilizar bootstrap em modelagem, avaliação de modelos e etc;3 - como não jogar informações fora com técnicas de modelagem ruins;4 - a importância de criar modelos interpretáveis ao invés de caixas-preta;e muito e muito mais. Esse deveria ser um livro de cabeceira para todo estatístico aplicado ou cientista de dados.Tem algum ponto ruim do livro? O autor é um bioestatístico e a maioria do exemplos são dessa área e com esse viés. Para mim não é de fato um problema mas pode ser algo relevante para alguns.

Un clásico en el análisis de la regresión. El autor desarrolla aquí su filosofía metodológica (a favor de la “regresión-regresión” en contraste con otras metodologías de “clasificación” y “black box” tan en boga en machine learning). Es muy pedagógico y apto para gente con distinta formación matemática

I purchased this book both for the study of data science and addressing a specific regression problem at work. Unlike more introductory textbooks, this book explains more practicalities as to which tests or techniques are better suited for the situation. It helps to know better what to do when the data is not ideal.

Very nice, extremely useful knowledge for people who want to understand regression modelling, easy to understand and has good examples. Hopefully there will be a newer version coming up.