Scikit Learn - Bernoulli Naïve Bayes

Bernoulli Naïve Bayes is another useful naïve Bayes model. The assumption in this model is that the features binary (0s and 1s) in nature. An application of Bernoulli Naïve Bayes classification is Text classification with ‘bag of words’ model. The Scikit-learn provides sklearn.naive_bayes.BernoulliNB to implement the Gaussian Naïve Bayes algorithm for classification.

Parameters

Following table consist the parameters used by sklearn.naive_bayes.BernoulliNB method −

Sr.No Parameter & Description
1

alpha − float, optional, default = 1.0

It represents the additive smoothing parameter. If you choose 0 as its value, then there will be no smoothing.
2

binarize − float or None, optional, default = 0.0

With this parameter we can set the threshold for binarizing of sample features. Binarization here means mapping to the Booleans. If you choose its value to be None it means input consists of binary vectors.
3

fit_prior − Boolean, optional, default = true

It tells the model that whether to learn class prior probabilities or not. The default value is True but if set to False, the algorithms will use a uniform prior.
4

class_prior − array-like, size(n_classes,), optional, Default = None

This parameter represents the prior probabilities of each class.

Attributes

Following table consist the attributes used by sklearn.naive_bayes.BernoulliNB method −

Sr.No Attributes & Description
1

class_log_prior_ − array, shape(n_classes,)

It provides the smoothed log probability for every class.
2

class_count_ − array, shape(n_classes,)

It provides the actual number of training samples encountered for each class.
3

feature_log_prob_ − array, shape (n_classes, n_features)

It gives the empirical log probability of features given a class $P\left(\begin{array}{c} features\arrowvert Y\end{array}\right)$.
4

feature_count_ − array, shape (n_classes, n_features)

It provides the actual number of training samples encountered for each (class,feature).

The methods of sklearn.naive_bayes.BernoulliNB are same as we have used in sklearn.naive_bayes.GaussianNB.

Implementation Example

The Python script below will use sklearn.naive_bayes.BernoulliNB method to construct Bernoulli Naïve Bayes Classifier from our data set −

import numpy as np
X = np.random.randint(10, size = (10, 1000))
y = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10])
from sklearn.naive_bayes import BernoulliNB
BNBclf = BernoulliNB()
BNBclf.fit(X, y)

Output

BernoulliNB(alpha = 1.0, binarize = 0.0, class_prior = None, fit_prior = True)

Now, once fitted we can predict the new value by using predict() method as follows −

Example

print((BNBclf.predict(X[0:5]))

Output

[1 2 3 4 5]
Advertisements