Mastering Predictive Analytics With R: Master the Craft of Predictive Modeling by Developing Strategy, Intuition, and a Solid Foundation in Essential Concepts

Into chapter 2 right now, so I will update review later. But so far so good. This book is the for the following audience:*have some experience with R, at least understand basic code, but not much more needed*have some experience with statistics. this isn't entry level, but its not advanced either. I've take a few analytics courses, and a good amount of this is review, but it serves well as a refresher and bringing a little more to the table.*want to get introduced to decision tree analysis and other models outside of regression (MNL, multivariable regression etc).*want to learn a little more about R and get introduced to good R packages for predictive analytics.Overall good buy so far, I am not disappointed

The choice of subjects and of examples is excellent, but this text suffers from very poor explanations. Much of the text is a kind of summary and rewriting of some chapters of the great book "Introduction to Statistical Learning" (ISL). But it isn't done consistenly well. The chapter on linear regression has good explanations, although not stellar (the author doesn't seem to care about using partial derivatives from calculus without first mentioning to readers about the advanced treatment). The chapters on logistic regression, neural networks and support vector machines have very poor explanations and suffer from the lack of a deeper conceptual framework and understanding. The chapter on decision tree has such a poor text that it makes a simple model look rather complicated. Most of the other chapters suffer from the same, especially the chapter on time series analysis that is also very poor.The examples in the text are interesting. The choice of the package caret for data preprocessing makes the code much more readable. The constant assessment tests are another great aspect of this book. But the hidden placement of cross-validation inside the chapter on support vector machines isn't a good idea. Cross-validation merits an earlier and more relevant treatment. This is an area where the author doesn't follow ISL that has a chapter specially devoted to resampling methods where cross-validation receives a deserved, distinguished treatment. The order of the chapters isn't good either. Neural networks and support vector machines should have come later, in any case after decision trees (as ISL did too). Apart from the poor and verbose explanations, it seems that this book also suffers where it departed from ISL, that is, where it treats subjects not covered there (neural networks and time series analysis, for example). Its explanations are sincerely bad and such a pain that its great examples aren't enough to sustain them.

Thorough discussion to predictive analytics methods with interesting examples from real life. Although the author sometimes seems to be assuming high knowledge of R, however he have shown efficient use of R in his examples.All in all, I believe it is a great book that any data scientist or predictive analytics practitioner should read and keep.

Now I'm not a fan of everything that Packt puts out; they've got some titles that really lacked qualify. However, I can say this is not one of them. Most of the 'Data Science Crack' books don't clearly explain topics like bagging and boosting. They don't go nearly deep enough into concepts like regression, and ASE, LASSO AND feature selection. These are NOT arbitrary and are not intuitive. As your typical IT / SQL DBA going into real Data Science, I can honestly say this book is a solid start. The 2nd edition looks promising as well, filling in a few gaps around deep learning.To answer another question, why R [for me at least]. I know SQL really well, and now w/ Microsoft SQL 2016 and higher, you can very easily integrate R models into the SQL code, thus you SQL server can parallelize your R models over an existing database [which has been 90+ % of the data I work with] using the RevoScaleR packages. It's been an easier transition for me given my background. Esp since R is very function-based, packages like dplyr and tidyverse felt fairly familiar to me.IMO I would also say that "Data Science" is not the same thing as "Deep Learning" I'm not doing image recognition or building robots. I'm working w/ very large data warehouses and helping both SQL DBA's and Business Analysts really up their game in Retail, Financial Services, Supply Chain / Distribution, and a bit of biotech [which is more SAS than Python & R combined :-) ]

I wish getting the data sets the books refer to was easier with the Kindle version.

well , I like this book . It teach me the basic of modelling as I am shifting from simple programmer to machine learning

Not worth for the money. Seems like author wrote it just register himself in Machine Learning domain.