GaussianNB

Algorithm overview

GaussianNB is a naive Bayes classifier for continuous numeric features. It models each feature per class with a Gaussian distribution and predicts the class with the highest posterior score.

JavaScript implementation

@kanaries/ml exposes Bayes.GaussianNB for lightweight probabilistic classification in browser or Node.js applications.

Quick start example

import { Bayes } from '@kanaries/ml';

const clf = new Bayes.GaussianNB({ varSmoothing: 1e-9 });
clf.fit([[1.0, 2.0], [1.2, 1.9], [4.0, 4.2], [4.2, 4.0]], [0, 0, 1, 1]);
const pred = clf.predict([[1.1, 2.1], [4.1, 4.1]]);

Detailed API reference

new Bayes.GaussianNB(props?: {
  priors?: number[] | null;
  varSmoothing?: number;
})

Methods:

• fit(X: number[][], y: number[]): void
• predict(X: number[][]): number[]

priors, when provided, must match the number of fitted classes.