Engineering features for modeling
As part of the system's understanding you would have gained some insights into your problem and dataset that can be used to create new features in your dataset by combining the existing features in various ways. For example, we can create a new feature called volume
by multiplying length, width, and height. Similarly, we can create a feature called mpg-ratio
by dividing highway-mpg
by city-mpg
. Let's also create a feature called cylinder-size
by dividing engine-size
by cylinder-count
. The equations for these features are as follows:
volume = length * width * height
mpg-ratio = highway-mpg / city-mpg
cylinder-size = engine-size / cylinder-count
Figure 4.11 shows an example of what these feature values look like:
As you can now see, many possibilities exist to create new features that could prove helpful in solving your problem...