#### Overview of this book

Feature engineering, the process of transforming variables and creating features, albeit time-consuming, ensures that your machine learning models perform seamlessly. This second edition of Python Feature Engineering Cookbook will take the struggle out of feature engineering by showing you how to use open source Python libraries to accelerate the process via a plethora of practical, hands-on recipes. This updated edition begins by addressing fundamental data challenges such as missing data and categorical values, before moving on to strategies for dealing with skewed distributions and outliers. The concluding chapters show you how to develop new features from various types of data, including text, time series, and relational databases. With the help of numerous open source Python libraries, you'll learn how to implement each feature engineering method in a performant, reproducible, and elegant manner. By the end of this Python book, you will have the tools and expertise needed to confidently build end-to-end and reproducible feature engineering pipelines that can be deployed into production.
Preface
Chapter 3: Transforming Numerical Variables
Chapter 4: Performing Variable Discretization
Chapter 5: Working with Outliers
Chapter 6: Extracting Features from Date and Time Variables
Chapter 7: Performing Feature Scaling
Chapter 8: Creating New Features
Chapter 9: Extracting Features from Relational Data with Featuretools
Chapter 10: Creating Features from a Time Series with tsfresh
Chapter 11: Extracting Features from Text Variables
Index
Other Books You May Enjoy

# Using power transformations

Power functions are mathematical transformations that follow , where lambda can take any value. The square and cube root transformations are special cases of power transformations where lambda is 1/2 or 1/3, respectively. The challenge resides in finding the value for the lambda parameter. The Box-Cox transformation, which is a generalization of the power transformations, finds the optimal lambda via maximum likelihood. We will discuss the Box-Cox transformation in the following recipe. In practice, we try different lambdas and visually inspect the variable distribution to determine which one offers the best transformation. In general, if the data is right-skewed – that is, observations accumulate toward lower values – we use lambda <1, while if the data is left-skewed – that is, there are more observations around higher values – then we use lambda >1.

In this recipe, we will carry out power transformations using NumPy...