Implementation:Online ml River NaiveBayes BernoulliNB

Knowledge Sources	Online_ml_River
Domains	Online_Learning, Naive_Bayes, Text_Classification
Last Updated	2026-02-08 16:00 GMT

Overview

Bernoulli Naive Bayes classifies based on binary feature occurrences, particularly effective for text classification with word presence/absence.

Description

This implementation treats features as binary variables based on a threshold. During training, it tracks whether each feature value exceeds true_threshold for each class. Predictions use the Bernoulli model: P(feature|class) for present features and 1-P(feature|class) for absent features. Laplace smoothing (alpha parameter) prevents zero probabilities. The model maintains class counts and feature occurrence counts, computing joint log-likelihood by summing log probabilities across all known features (not just those present in the query).

Usage

Use Bernoulli NB for text classification with bag-of-words features when word presence matters more than frequency. It's particularly suited for short documents or when vocabulary size is large. Works well with binary or count features after thresholding. Supports both single-instance and mini-batch learning. Pair with feature extraction like BagOfWords for text data.

Code Reference

Source Location

Repository: Online_ml_River
File: river/naive_bayes/bernoulli.py

Signature

class BernoulliNB(base.BaseNB):
    def __init__(self, alpha=1.0, true_threshold=0.0):
        self.alpha = alpha
        self.true_threshold = true_threshold
        self.class_counts = collections.Counter()
        self.feature_counts = collections.defaultdict(collections.Counter)

Import

from river import naive_bayes

I/O Contract

Parameters

Parameter	Type	Default	Description
alpha	float	1.0	Laplace/Lidstone smoothing parameter
true_threshold	float	0.0	Threshold for binarizing features

Attributes

Attribute	Type	Description
class_counts	Counter	Number of instances per class
feature_counts	defaultdict	Feature occurrence counts per class

Input/Output

Method	Input	Output
learn_one	x: dict, y: Any	None
learn_many	X: DataFrame, y: Series	None
predict_proba_one	x: dict	dict
predict_proba_many	X: DataFrame	DataFrame

Usage Examples

import pandas as pd
from river import compose
from river import feature_extraction
from river import naive_bayes

docs = [
    ("Chinese Beijing Chinese", "yes"),
    ("Chinese Chinese Shanghai", "yes"),
    ("Chinese Macao", "yes"),
    ("Tokyo Japan Chinese", "no")
]

# Single instance learning
model = compose.Pipeline(
    ("tokenize", feature_extraction.BagOfWords(lowercase=False)),
    ("nb", naive_bayes.BernoulliNB(alpha=1))
)

for sentence, label in docs:
    model.learn_one(sentence, label)

model["nb"].p_class("yes")
# 0.75
model["nb"].p_class("no")
# 0.25

model.predict_proba_one("test")
# {'yes': 0.883..., 'no': 0.116...}

model.predict_one("test")
# 'yes'

# Mini-batch learning
X = pd.Series([
   "Chinese Beijing Chinese",
   "Chinese Chinese Shanghai",
   "Chinese Macao",
   "Tokyo Japan Chinese"
])

y = pd.Series(["yes", "yes", "yes", "no"])

model = compose.Pipeline(
    ("tokenize", feature_extraction.BagOfWords(lowercase=False)),
    ("nb", naive_bayes.BernoulliNB(alpha=1))
)

model.learn_many(X, y)

unseen = pd.Series(["Taiwanese Taipei", "Chinese Shanghai"])

model.predict_proba_many(unseen)
#          no       yes
# 0  0.116846  0.883154
# 1  0.047269  0.952731

model.predict_many(unseen)
# 0    yes
# 1    yes
# dtype: object

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment