Existing variables can be combined to create new insightful features. We discussed how to combine variables using common mathematical and statistical operations in the previous two recipes, Combining multiple features with statistical operations and Combining pairs of features with mathematical functions. A combination of one feature with itself, that is, a polynomial combination of the same feature, can also be quite informative or increase the predictive power of our algorithms. For example, in cases where the target follows a quadratic relationship with a variable, creating a second degree polynomial of the feature allows us to use it in a linear model, as shown in the following screenshot:
In the plot on the left, due to the quadratic relationship between the target, y, and the variable, x, there is a poor linear fit. Yet, in the...
