Applied Predictive Modeling Author: Max Kuhn | Language: English | ISBN:
B00CU4Q4QS | Format: PDF
Applied Predictive Modeling Description
This text is intended for a broad audience as both an introduction to predictive models as well as a guide to applying them. Non-mathematical readers will appreciate the intuitive explanations of the techniques while an emphasis on problem-solving with real data across a wide variety of applications will aid practitioners who wish to extend their expertise. Readers should have knowledge of basic statistical ideas, such as correlation and linear regression analysis. While the text is biased against complex equations, a mathematical background is needed for advanced topics.
- File Size: 7141 KB
- Print Length: 620 pages
- Publisher: Springer; 2013 edition (May 17, 2013)
- Sold by: Amazon Digital Services, Inc.
- Language: English
- ASIN: B00CU4Q4QS
- Text-to-Speech: Enabled
X-Ray:
- Lending: Not Enabled
- Amazon Best Sellers Rank: #110,824 Paid in Kindle Store (See Top 100 Paid in Kindle Store)
- #4
in Kindle Store > Kindle eBooks > Nonfiction > Professional & Technical > Professional Science > Biological Sciences > Biostatistics - #11
in Kindle Store > Kindle eBooks > Nonfiction > Professional & Technical > Medical eBooks > Research - #52
in Kindle Store > Kindle eBooks > Nonfiction > Science > Mathematics > Applied > Probability & Statistics
- #4
in Kindle Store > Kindle eBooks > Nonfiction > Professional & Technical > Professional Science > Biological Sciences > Biostatistics - #11
in Kindle Store > Kindle eBooks > Nonfiction > Professional & Technical > Medical eBooks > Research - #52
in Kindle Store > Kindle eBooks > Nonfiction > Science > Mathematics > Applied > Probability & Statistics
I read "Applied predictive modeling" (which I will shorten to APM) shortly after I read "Introduction to statistical learning" (ISL) by James, Witten, Hastie and Tibshirani, and find that book both closest to APM, and helpful in highlighting APM's strengths.
The two books cover the same broad subject. If you google "kuhn caret", you will find Max Kuhn's (very informative) presentation of his "caret" R package, and its first slide will tell you that he uses "predictive modeling" as a synonym of "machine learning" - what Hastie and Tibshirani call "statistical learning". Adopting H&T's terminology choice, I will say that both books combine theory of "statistical learning" with hands-on illustrations and exercises implemented in R; the get-your-hands-dirty, try-it-out element is, in fact, ISL's key difference from the earlier, venerable "Elements of statistical learning".
Both books, inevitably, go over a catalog of statistical-learning techniques. The shorter ISL, in my opinion, is superior at explaining the concepts and communicating the principles, while APM takes the more straightforward approach of "beefing up" the catalog, by spending more pages on each item and including more items. While ISL is by design very accessible, APM can be more technical - the detail will surely be appreciated by any practitioner - and, as it talks about the various methods, it can and does discuss recent extensions, offering an extensive and "fresh" bibliography. R-wise, APM's advantage is not decisive (if you look at content, not line count) but big; the book naturally favors "caret" - which has a useful role, "wrapping" a plethora of third-party R packages, and providing a common interface, plus helpful utilities - but both references and uses the specialist packages as well.
There are many fine math-oriented predictive modeling books, such as Hastie (The Elements of Statistical Learning: Data Mining, Inference, and Prediction, Second Edition (Springer Series in Statistics)). Kuhn et al consider them "sister texts" and begin immediately to differentiate-- their approach is hands on and practical, for the express purpose of demonstrating HOW to sort, structure and predict via Python or R, for the purpose of accuracy and understanding of the DATA and trends, NOT learning the underlying math.
For a couple of pharmaceutical guys, (who BTW use R extensively, I've been an analyst in that industry), you'd think the examples would be new chemical or biological entities. Not so! The cases are fun and exciting, ranging from the nontrivial compression strength of concrete (want that bridge to hold when you cross?) to fuel economy, credit scoring, success in grant applications (boy their colleagues will love that one!), and cognitive impairment. I evaluate technology for patents at payroy dot com, and we have a log likelihood model using Bayesian and Monte Carlo that their grant section helped translate seamlessly to R! We're NOT talking pie in the sky pseudo code here, but real life, real results recipes.
The authors talk about the "scholarly veil" -- meaning we general workers and researchers don't always "deserve" to see the underlying process, software and data (and, other than open source, often can't afford it). Wow, do they pop that myth!
Applied Predictive Modeling Preview
Link
Please Wait...