Evaluarea Out Of Bag

Machine Learning cu modele bazate pe arbori în Python

Elie Kawerk

Data Scientist

Bagging

unele instanțe pot fi eșantionate de mai multe ori pentru un model,
alte instanțe pot să nu fie eșantionate deloc.

Instanțe Out Of Bag (OOB)

În medie, pentru fiecare model, 63% din instanțele de antrenament sunt eșantionate.
Restul de 37% constituie instanțele OOB.

Evaluare OOB

evaluare oob

Evaluare OOB în sklearn (Setul de date Breast Cancer)

# Import models and split utility function
from sklearn.ensemble import BaggingClassifier
from sklearn.tree import DecisionTreeClassifier
from sklearn.metrics import accuracy_score
from sklearn.model_selection import train_test_split

# Set seed for reproducibility
SEED = 1

# Split data into 70% train and 30% test
X_train, X_test, y_train, y_test =  train_test_split(X, y, test_size= 0.3,
                                                     stratify= y,
                                                     random_state=SEED)

# Instantiate a classification-tree 'dt'
dt = DecisionTreeClassifier(max_depth=4,
                            min_samples_leaf=0.16,
                            random_state=SEED)

# Instantiate a BaggingClassifier 'bc'; set oob_score = True 
bc = BaggingClassifier(base_estimator=dt, n_estimators=300,
                       oob_score=True, n_jobs=-1)


# Fit 'bc' to the training set
bc.fit(X_train, y_train)

# Predict the test set labels
y_pred = bc.predict(X_test)

# Evaluate test set accuracy
test_accuracy = accuracy_score(y_test, y_pred)

# Extract the OOB accuracy from 'bc'
oob_accuracy = bc.oob_score_

# Print test set accuracy
print('Test set accuracy: {:.3f}'.format(test_accuracy))

Test set accuracy: 0.936

# Print OOB accuracy
print('OOB accuracy: {:.3f}'.format(oob_accuracy))

OOB accuracy: 0.925

Să exersăm!

Machine Learning cu modele bazate pe arbori în Python