Automatic Dataset Normalization for Feature Engineering in Python


A normalized, relational dataset makes it easier to perform feature engineering. Unfortunately, raw data for machine learning is often stored as a single table, which makes the normalization process tedious and time-consuming. Well, I am happy to introduce you to AutoNormalize, an open-source library that automates the normalization process and integrates seamlessly with Featuretools, another open-source library for automated feature engineering. The normalized dataset can then be returned as either an EntitySet or a collection of DataFrames. Using AutoNormalize makes it easier to get started with Featuretools and can help provide you with a quick preview of what Featuretools is capable of.