Building Machine Learning Systems with Python - Third Edition

Looking at the decision boundaries

We will now examine the decision boundaryies. In order to plot these on paper, we will simplify and look at only two dimensions:

knn.fit(features[:, [0,2]], target)  

We will call predict on a grid of feature values (1000 by 1000 points):

y0, y1 = features[:, 2].min() * .9, features[:, 2].max() * 1.1 
x0, x1 = features[:, 0].min() * .9, features[:, 0].max() * 1.1 
X = np.linspace(x0, x1, 1000) 
Y = np.linspace(y0, y1, 1000) 
X, Y = np.meshgrid(X, Y) 
C = knn.predict(np.vstack([X.ravel(), Y.ravel()]).T).reshape(X.shape)  

Now, we plot the decision boundaries:

cmap = ListedColormap([(1., 1., 1.), (.2, .2, .2), (.6, .6, .6)]) fig,ax = plt.subplots() ax.scatter(features[:, 0], features[:, 2], c=target, cmap=cmap) for ...