Machine Learning with Swift

Understanding model capacity trade-offs

Let's train trees with different depths: starting from 1 split, and to maximal 23 splits:

In []: 
train_losses = [] 
test_losses = [] 
for depth in xrange(1, 23): 
    tree_model.max_depth = depth 
    tree_model = tree_model.fit(X_train, y_train) 
    train_losses.append(1 - tree_model.score(X_train, y_train)) 
    test_losses.append(1 - tree_model.score(X_test, y_test)) 
figure = plt.figure()  
plt.plot(train_losses, label="training loss", linestyle='--') 
plt.plot(test_losses, label="test loss") 
plt.legend(bbox_to_anchor=(0., 1.02, 1., .102), loc=3, ncol=2, mode="expand", borderaxespad=0.) 
Out[]: 

Figure 2.8: Training loss ...