Decision Trees

Machine Learning for Finance in Python

Nathan George

Data Science Professor

Decision trees

basic decision tree

Decision trees

decision tree with targets

Decision tree splits

categorical decision tree split

Decision tree splits

numerical decision tree split

Bad tree

bad regression split

Good tree

good regression split

Decision tree regression

decision tree predictions

Regression trees

from sklearn.tree import DecisionTreeRegressor

decision_tree = DecisionTreeRegressor(max_depth=5)

decision_tree.fit(train_features, train_targets)

Decision tree hyperparameters

max depth of trees

Max depth of 3

max depth of 3 decision tree

Evaluate model

print(decision_tree.score(train_features, train_targets))
print(decision_tree.score(test_features, test_targets))

0.6662215501032416
-0.08917300191734268

train_predictions = decision_tree.predict(train_features)
test_predictions = decision_tree.predict(test_features)
plt.scatter(train_predictions, train_targets, label='train')
plt.scatter(test_predictions, test_targets, label='test')
plt.legend()
plt.show()

predictions vs actual

Grow some trees!

Machine Learning for Finance in Python