Hands-On Automated Machine Learning

Principal Component Analysis in action

The following code block shows you how to apply PCA with two components and visualize the results:

# PCAfrom sklearn.decomposition import PCApca = PCA(n_components=2, whiten=True)pca = pca.fit_transform(df)plt.scatter(pca[:, 0], pca[:, 1], c=data.target, cmap="RdBu_r", edgecolor="Red", alpha=0.35)plt.colorbar()plt.title('PCA, n_components=2')plt.show()

We get the following plot from the preceding code:

Plot of PCA, n_components=2

Here, you can see the red class (dark gray) is very condensed into one particular area and it's hard to separate classes. Differences in variances distort our view and scaling ...