Python Feature Engineering Cookbook covers well-demonstrated recipes focused on solutions that will assist machine learning teams in identifying and extracting features to develop highly optimized and enriched machine learning models. This book includes recipes to extract and transform features from structured datasets, time series, transactions data and text. It includes recipes concerned with automating the feature engineering process, along with the widest arsenal of tools for categorical variable encoding, missing data imputation and variable discretization. Further, it provides different strategies of feature transformation, such as Box-Cox transform and other mathematical operations and includes the use of decision trees to combine existing features into new ones. Each of these recipes is demonstrated in practical terms with the help of NumPy, SciPy, pandas, scikit-learn, Featuretools and Feature-engine in Python.
Throughout this book, you will be practicing feature generation, feature extraction and transformation, leveraging the power of scikit-learn’s feature engineering arsenal, Featuretools and Feature-engine using Python and its powerful libraries.
