My Review on Feature Engineering Book "Python Feature Engineering Cookbook"

?? Feature Engineering is heart of machine learning which is also known as Applied Machine learning. I was pleased to find that One Amazing Book Name "Python Feature Engineering Cookbook: Over 70 recipes for creating, engineering, and transforming features to build machine learning models"

Amazon : https://www.amazon.in/Python-Feature-Engineering-Cookbook-transforming/dp/1789806313/

???? This book is quite good for Applied feature engineering practitioners.

?

Disclaimer : The publisher requested that I review this book and gave me a review copy. I am assured to be 100% genuine about this book, I feel both good and less about this book.

Overview:

This book is for those who have good background knowledge of machine learning and explore to generate more feature engineering techniques. It gives a variety of ways to articulate the new feature with a creative approach. It comprises Basic data preparation, feature transformation, feature selection to feature engineering techniques.

What I like:

This book did a wonderful job of covering the major topics that data science professionals use frequently. The first chapter explained very well how to examine the data set and understand the different characteristics of features and how to transform them. The second chapter gives a good idea of how to deal with missing value and how to deal with the help of different techniques.

Chapter three, which is most important from my point of view, because it explains how to handle the categorical variable in different situations and which is best encoding method for particular type of feature. Chapters five and six contain much more information about the spreadness of the data and how to deal with outliers.

The final chapters discuss best practices for generating the time based features, performing feature scaling, and using mathematical techniques to generate the new features in existing data.

Most of the time I like all the last chapters better because it's more about the creative approach of presenting new feature generations.

What I not like:

One thing I want to draw your attention toward is the need for complex data to explain the example so that users can learn more about details.

In the last few chapters you used time-based basic data. But in the real data set it's not as easy as it is here in these books.

What I would like to see:

There is only little I would change in this book and that is examples. Each example needs some complexity and showcase like in such a situation where you need to focus. This is most important from my point of view because the industry datasets is a bit more complex than that shown in the book.

I must suggest this book to those who wants to become master of feature engineering.