Naive Bayes Predict

Problem

Predict a class with Gaussian naive Bayes for a single point x.

classes, priors, means, and variances are parallel per-class lists; means[ci][j] and variances[ci][j] are the mean and variance of feature j for class ci. Score each class as log(prior) plus the sum of log Gaussian densities, then return the class with the highest score, breaking ties by the earlier entry in classes.

Hint

Work in log space: log prior plus the sum of per-feature log densities.

Test cases

case 1

in: [[5.1,3.5],["A","B"],[0.5,0.5],[[5,3.4],[6,2.8]],[[0.04,0.09],[0.25,0.16]]]

expected: "A"

case 2

in: [[6.2,3],["A","B"],[0.5,0.5],[[5,3.4],[6,2.8]],[[0.04,0.09],[0.25,0.16]]]

expected: "B"

case 3

in: [[5,3.4],["A","B"],[0.8,0.2],[[5,3.4],[6,2.8]],[[1,1],[1,1]]]

expected: "A"

case 4

in: [[0,0],["neg","pos"],[0.5,0.5],[[0,0],[5,5]],[[1,1],[1,1]]]

expected: "neg"

Starter codePython

import math
def naive_bayes_predict(x, classes, priors, means, variances):
    # Your code here
    pass

4 test cases · run your code against them in the editor