

In [ ]:

    
import numpy as np
import matplotlib.pyplot as plt
% matplotlib inline
plt.rcParams["figure.dpi"] = 200
np.set_printoptions(precision=3)
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import scale, StandardScaler



In [ ]:

    
from sklearn.datasets import load_breast_cancer
cancer = load_breast_cancer()



In [ ]:

    
X_train, X_test, y_train, y_test = train_test_split(
    cancer.data, cancer.target, stratify=cancer.target, random_state=0)

Random Forests



In [ ]:

    
from sklearn.ensemble import RandomForestClassifier
from sklearn.datasets import load_digits
digits = load_digits()
X_train, X_test, y_train, y_test = train_test_split(
    digits.data, digits.target, stratify=digits.target, random_state=0)



In [ ]:

    
train_scores = []
test_scores = []

rf = RandomForestClassifier(warm_start=True)
estimator_range = range(1, 100, 5)
for n_estimators in estimator_range:
    rf.n_estimators = n_estimators
    rf.fit(X_train, y_train)
    train_scores.append(rf.score(X_train, y_train))
    test_scores.append(rf.score(X_test, y_test))



In [ ]:

    
plt.plot(estimator_range, test_scores, label="test scores")
plt.plot(estimator_range, train_scores, label="train scores")
plt.ylabel("accuracy")
plt.xlabel("n_estimators")
plt.legend()



In [ ]:

    
# out of bag predictions



In [ ]:

    
train_scores = []
test_scores = []
oob_scores = []

feature_range = range(1, 64, 5)
for max_features in feature_range:
    rf = RandomForestClassifier(max_features=max_features, oob_score=True, n_estimators=200, random_state=0)
    rf.fit(X_train, y_train)
    train_scores.append(rf.score(X_train, y_train))
    test_scores.append(rf.score(X_test, y_test))
    oob_scores.append(rf.oob_score_)



In [ ]:

    
plt.plot(feature_range, test_scores, label="test scores")
plt.plot(feature_range, oob_scores, label="oob_scores scores")
plt.plot(feature_range, train_scores, label="train scores")
plt.legend()
plt.ylabel("accuracy")
plt.xlabel("max_features")



In [ ]:

    
X_train, X_test, y_train, y_test = train_test_split(
    cancer.data, cancer.target, stratify=cancer.target, random_state=1)
rf = RandomForestClassifier(n_estimators=100).fit(X_train, y_train)



In [ ]:

    
rf.feature_importances_



In [ ]:

    
pd.Series(rf.feature_importances_,
          index=cancer.feature_names).plot(kind="barh")

Exercise

Use a random forest classifier or random forest regressor on a dataset of your choice. Try different values of n_estimators and max_depth and see how they impact performance and runtime. Tune max_features with GridSearchCV.