Why is feature engineering considered the "dark art" of machine learning? Transforming raw data into a form that your machine learning algorithm can utilize seems mysterious and downright frightening! Bring your wizard hat and join me as this machine learning apprentice shares her personal book of feature engineering incantations.
What is feature engineering and why do we need it? When is it applied? Is it an art or a science? Find out the answers to these questions and more as we explore different methods of feature engineering with practical examples provided. There are three main methods of feature engineering: adjusting raw features, combining raw features and decomposing raw features into usable subsets. We will use datasets to illustrate binning, encoding, binaries, summing, differencing, feature scaling, extraction, and the manipulation of date/time features. Finally, we will explore the performance of a machine learning model before and after feature engineering is applied. As a postscript, current automated feature engineering tools for Python will be introduced.