Marketing Data Science: Modeling Techniques in Predictive Analytics with R and Python (FT Press Analytics) 1st Edition

Very good book, well written, and the best pas, as with all of Miller's books that I have purchased, is that it comes with real code examples in both Python and R. Great way to get up and running.

good book, but....no data sets to work with. Seems critical for a source code heavy book (ie almost every chapter has pages of code). We would prefer not to scan, then try to run the code ourselves. Read the appendix first at that seems to be where the theory is then go back to the chapters for practical work. Borrowed this book from the library....its really expensive otherwiseupdate: OK --kept reading, and paying overdue fines at the library, so I bought the book. Really worth the read if you're serious about focusing on marketing data science. Great starting read for technical marketers who want to do something in this field.Glad the code samples are now available. Will have to test and see. One thing I noticed about the content. Every time something got interesting Prof. Miller would quote a reference for further reading (ie. details). That's sort of OK, but leaves me wanting and having to go dig elsewhere. Suggestion: one more paragraph for such situations would put my curiosity at rest. A lot of content around product development, positioning, recommending, but a little light on broader examples - It might be helpful to describe a broader range of techniques (ie. list them), then drill down on one or two. It just seems too narrow, like drinking from a straw when really a funnel is needed with the huge alternatives. Enjoyed the book (looks like a text book but reads like a novel - that's a good thing)

Updated my rating from 2 to 5 stars as the code has become available on FTPress. I received an email from the publisher last week. Not sure why it took over 6 months for them to post this.

Good book, very well explained examples (the R and the Python codes are very well written) but if you have read other books from Prof. Miller, you would be able to remember some exacts paragraphs across some books.

If you want to have just one book on Marketing Data Science, this is the one.

First, this is an excellent book. Why 4 stars? It presupposes a fair amount of knowledge of R, Python and the analytic techniques used for the data science approach of marketing analytics. Newbies will struggle with the concepts in this book. It dives right into the code, the techniques, and the statistical measures used in the first example of how Miller made the decision of which mobile company to use after a move based on 16 product profiles using linear models and conjoint analysis. (a brief explanation of the sum contrasts and fitted regression coefficient follows with code for R and Python). In many of the chapters, there are appendices that have additional descriptions of statistical techniques and references.That being said, this is an excellent reference for programmers, statisticians, modelers, marketers and data scientists. Props to Miller for calling out in the intro the definition of data science, "the new statistics, a blending of modeling techniques, information technology, and business savvy" while also reminding readers that it doesn't mean abandoning scientific research, surveys, inference, or primary data collection. Not all of the information that we need or want is available.The pros: readers familiar with how to program in R or Python will be able to follow along. All the sample code tested have been correct so far. Python scripts are notoriously problematic in books because of the spacing requirements, but all those tested have worked. Since it's focused on using Python and R (both of which are free and have robust user communities), people who are familiar with the analytic techniques but who want to learn to apply it can do so without having to pay for software.Miller also includes concepts of data visualizations (often done through R, which has a number of great packages he references), data base systems and data preparation, all of which are major components of being able to efficiently answer the question. Data curation is often a huge undertaking that non-database people misunderstand.To get a sense of the comprehensiveness of the appendices, in the marketing data source appendix B, Miller also describes options for sampling (simple random, convenience, probability samples, types of sampling frame, cluster sampling, adaptive sampling, on-line surveys, telephone surveys, random-digit dial, and discusses some of the pros and cons of each (with words, not formulas). He also discusses the concept of a panel sample and the accuracy of the panel in being able to generalize the responses back to the population. The one discussion area that is missed when he asks " How large a ample is needed to obtain useful information?" is a power calculation - looking at what is the smallest needed sample to detect a difference in what measure is being used (improvement on an indicator/survey question; differences between specific populations; or to determine how wide a standard error is around the point estimate to see what a detectable difference would be.Overall, though, it's got a lot of great information. It's dense and takes time to work through the examples and code (especially for someone like me who is just an R and Python tinkerer without a ton of expertise in those languages), but is a good primer that presents the concepts in a neutral way, allowing the data scientist to make up their own mind about how to approach the situation.

I admit to struggling with programming languages. Over the years, I’ve learned a few, but still consider myself to be a brute-force coder. Fortunately, with the growth of free-lance brokering services, over the past few years, I’ve been able to procure third-parties to write the code I need at reasonable rates.Python, however, is a different story. I want to be personally proficient in Python for a number of reasons. I started hearing about R a year or so ago and have read one very entertaining and informative book on using it to analyze baseball statistics.So I came to Professor Miller’s book with high hopes and I am pleased to say they have been fulfilled.The second page of his Preface, Miller provides the clearest explanation of the relationship between Python and R that I have seen. It cleared up several points of confusion and questioning for me.Miller’s writing style is delightfully light and familiar, eschewing the dead hand style of most academic work.He leaps right in explaining in very easy to understand language the nature of models and modeling.Yes, things do grow more complex as Miller explores his subject over the next 300 pages.Data examples are downloadable and the book itself is filled with clear illustrations of Miller’s points.I still need to work more on pattern matching and Regular Expressions in Python before I can move into serious study of this book, but from my initial examination, it is certainly one I want to spend considerable time with. Just page 2 of the Preface makes this worthwhile for me.Jerry