With his latest book on marketing data science, Thomas Miller continues his series of books on applied data mining and modeling with R and Python. I’ve read his previous book [1] and was pleasantly surprised to rediscover the same pragmatic approach, self-contained writing style, and ready-to-use code snippets here.
Marketing data science shares a common background with predictive analysis but aims more specifically at understanding the markets, and business-to-consumer and business-to-business interactions. These interactions can be roughly classified into understanding the preferences of users, being able to predict their choices, retaining and targeting current customers, developing and positioning products, predicting sales, and recommending new products.
Business intelligence and social media analytics for sentiment analytics are also part of the overall strategy required for doing marketing research. The author addresses each of these topics and provides real-world illustrations, relevant marketing questions, data samples, and sample code in order to show how concrete cases can be approached and solved. The scientific component in performing these tasks relies on identifying the sound analytical framework and data processing environment. Though most of the analytical approaches rely on clustering techniques, plots, regression, and recommendation engines, the author makes a very good point in showing how each of these techniques can be deployed for marketing.
The second part of the book consists of three appendices. The first appendix is a condensed introduction to machine learning, data analytics, and further references to relevant academic work in the area. The following appendices discuss several marketing data sources and several case studies, including bank marketing, transportation systems, and Wikipedia votes. This part of the book could have been its own dedicated book, as the quality and quantity of information is extremely valuable and highly useful for both applied data analytics and graduate students in data analytics and machine learning. Some of the recipes and examples can be easily extended and reused for many common marketing-related tasks, and this is one of the strongest points of the book. While a pure marketing-related book might not be practical for developers, and a pure programming book is difficult for nonprogrammers, Miller manages to bridge the two worlds in a unified approach through hands-on, practice-driven content and realistic case studies, which are relevant for both a seasoned marketer and interested developers. Programmers with basic experience in a programming language will easily understand the standard Python examples (the particular version of the Python language is not important) and extend them for a similar project. For a nonprogrammer marketer, however, the code itself is not directly usable, but might be a good starting point to interface with the development team and better understand how typical marketing problems might be solved. Straightforward application of the code snippets will not be enough in an operational environment because many subtle details are not addressed in the book--integration with database servers, customer management systems, and poor data being the most relevant ones; these will need to be tackled in a real-world marketing scenario.
I highly recommend this book to any aspiring data analyst or marketing researcher. Its no-nonsense approach, relevant case studies, and deep insights into the fundamental solutions for market analysis make it an excellent handbook and tutorial in this area.
More reviews about this item: Amazon
