objectif

Naive Bayes multinomial pour comptages (bag-of-words).

code minimal

from sklearn.naive_bayes import MultinomialNB
from sklearn.feature_extraction.text import CountVectorizer
X = ["spam spam", "ham eggs", "spam eggs"]
y = [1,0,1]
V = CountVectorizer().fit_transform(X)
clf = MultinomialNB().fit(V, y)
print(clf.predict(V[:1])[0] in [0,1])

utilisation

print(clf.predict_proba(V[:1]).shape[1] == 2)

variante(s) utile(s)

from sklearn.naive_bayes import BernoulliNB
print(hasattr(BernoulliNB(), "fit"))

notes

• Adapté aux features non négatives (comptages).