Feature Engineering and Selection: A Practical Approach for Predictive Models

by Max Kuhn, Kjell Johnson

The process of developing predictive models includes many stages. Most resources focus on the modeling algorithms but neglect other critical aspects of the modeling process. This book describes techniques for finding the best representations of predictors for modeling and for nding the best subset of predictors for improving model performance. A variety of example data sets are used to illustrate the techniques along with R programs for reproducing the results.

Genres: Programming, Textbooks, Nonfiction

314 pages, Hardcover

Published August 2, 2019

Reviews

[Minervas Owl · Rating 3 out of 5]

The book can serve as a quick guide to specific feature engineering/selection techniques, but it's by no means systematic and thorough. Leaving aside the first few chapters on modeling basics and pre-processing ABC, the remaining 150 or so pages feel too light for the topics. For example, about feature selection, the authors mainly introduced what Kuhn has implemented in the R caret package: recursive feature elimination, simulated annealing, and the genetic algorithm. According to this article below, these are only a small part of the many methods available, https://www.mdpi.com/1999-5903/12/3/54 .

I benefited the most from the authors' emphasis on resampling and statistical testing. For example, how to check whether a variable significantly improves prediction accuracy such as AUC? One could permutate the variable randomly to generate a distribution of AUC, then compare it with the AUC generated by the real data.

[Francisco Lima · Rating 4 out of 5]

Great purchase, did not disappoint me. It clearly illustrates the usage of feature engineering techniques and makes the R code available from the GitHub repository accessible via the bit.ly links in the closing section from each chapter. I hope the many harmless typos will be corrected in future revisions.

[Pritesh Shrivastava · Rating 5 out of 5]

One of the best resources on ML out there, and it's completely free!

[Pawin · Rating 4 out of 5]

The book contains info about feature engineering and selection, and also basic detail about data mining process, such as modeling, preprocessing, and etc. While the topics seems to be simple, reader needs a lot of prior knowledge to be able to understand all contents.