Feature Engineering Made Easy

LDA versus PCA – iris dataset

Finally, we arrive at the moment where we can try using both PCA and LDA in our machine learning pipelines. Because we have been working with the iris dataset extensively in this chapter, we will continue to demonstrate the utility of both LDA and PCA as feature transformational pre-processing steps for supervised and unsupervised machine learning.

We will start with supervised machine learning and attempt to build a classifier to recognize the species of flower given the four quantitative flower traits:

1. We begin by importing three modules from scikit-learn:

from sklearn.neighbors import KNeighborsClassifierfrom sklearn.pipeline import Pipelinefrom sklearn.model_selection import cross_val_score

We will use ...