class NBTest(unittest.TestCase):
def setUp(self):
self.mnb = NaiveBayes(multinomial=True)
self.skmnb = MultinomialNB()
self.bnb = NaiveBayes(bernoulli=True)
self.skbnb = BernoulliNB()
self.cnb = NaiveBayes(multinomial=True, cnb=True)
self.wcnb = NaiveBayes(multinomial=True, wcnb=True)
def test_count_vectorized(self):
self.mnb.fit(X_count, train_targets)
self.skmnb.fit(X_count, train_targets)
self.assertEqual(self.mnb.score(X_count_test,test_targets),self.skmnb.score(X_count_test,test_targets))
def test_tfidf_vectorized(self):
self.mnb.fit(X_tfidf, train_targets)
self.skmnb.fit(X_tfidf, train_targets)
self.assertEqual(self.mnb.score(X_tfidf_test, test_targets), self.skmnb.score(X_tfidf_test, test_targets))
def test_cnb(self):
self.cnb.fit(X_count, train_targets)
self.mnb.fit(X_count, train_targets)
cnb_score = self.cnb.score(X_count_test, test_targets)
mnb_score = self.mnb.score(X_count_test, test_targets)
print "CNB: {}, MNB: {}".format(cnb_score, mnb_score)
assert (cnb_score - mnb_score) > -0.1
def test_wcnb(self):
self.wcnb.fit(X_count, train_targets)
self.mnb.fit(X_count, train_targets)
wcnb_score = self.wcnb.score(X_count_test, test_targets)
mnb_score = self.mnb.score(X_count_test, test_targets)
print "WCNB: {}, MNB: {}".format(wcnb_score, mnb_score)
assert (wcnb_score - mnb_score) > -0.5
"""Prediction on the data set"""
import pandas as pd
from classifiers.naive_bayes import NaiveBayes
from classifiers.util import train_test_split, confusion_matrix, accuracy_score
# reading the data set
df = pd.read_csv("./dataset/processed.csv")
X = df.drop("Outcome", axis=1).values
y = df["Outcome"].values
# making sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=20)
# training the model
nb = NaiveBayes(X_train, y_train)
nb.fit(X_train, y_train)
# getting the predictions
y_predictions = nb.predict(X_test)
print(f"The accuracy score :: {accuracy_score(y_predictions, y_test) * 100} %")
# confusion matrix
print("Confusion Matrix ::")
print(f"{confusion_matrix(y_test, y_predictions)}")
