Creating new features

Feature creation and feature selection are can be some of the most time-consuming part of model-building. It is often the place where the interplay between technical knowledge (for example about the models we intend to build) and business or domain knowledge (data requirements and how the analysis will be put to use) is at the forefront.

Let's review our workflow so far:

1. load all the data (and combine them if necessary)
2. inspect the data in preparation cleaning it
3. clean the data in preparation for analysis
4. add any interesting features or columns as far as they pertain to the analysis
5. find ways to analyze or summarize the data and report your findings