naive_bayes_toy_example.py

import numpy as np


# data
heights = np.array([6.3, 5.9, 5.1, 5.6, 5.1])
weights = np.array([50.2, 79.7, 61.4, 47.1, 59.8])
classes = np.array(["Male", "Female", "Female" , "Male", "Female"])

# P(Class)
males_count = 0
females_count = 0
sample_size = len(classes)

for x in classes:
  if x == "Male":
    males_count += 1
  else:
    females_count += 1

p_male = males_count / sample_size
p_female = females_count / sample_size

# PDF
heights_of_males = []
weights_of_males = []
heights_of_females = []
weights_of_females = []

for i in range(sample_size):
  if classes[i] == "Male":
    heights_of_males.append(heights[i])
    weights_of_males.append(weights[i])
  else:
    heights_of_females.append(heights[i])
    weights_of_females.append(weights[i])

mean_height_males = np.mean(heights_of_males)
mean_weight_males = np.mean(weights_of_males)
mean_height_females = np.mean(heights_of_females)
mean_weight_females = np.mean(weights_of_females)
var_height_males = np.var(heights_of_males)
var_weight_males = np.var(weights_of_males)
var_height_females = np.var(heights_of_females)
var_weight_females = np.var(weights_of_females)


def the_pdf(x, mean, variance):
  pd = 1 / (np.sqrt(2 * np.pi * variance)) * np.exp((-(x - mean)**2) / (2 * variance))
  return pd

# Predict
x = [5.8, 82.1] # [height, weight]

p_height_male = the_pdf(x[0], mean_height_males, var_height_males)
p_weight_male = the_pdf(x[1], mean_weight_males, var_weight_males)
p_height_female = the_pdf(x[0], mean_height_females, var_height_females)
p_weight_female = the_pdf(x[1], mean_weight_females, var_weight_females)

# Get class probabilities
p_male_h_and_w = p_male * p_height_male * p_weight_male
p_female_h_and_w = p_female * p_height_female * p_weight_female
print("P(Male | height & weight) =", p_male_h_and_w)
print("P(Female | height & weight) =", p_female_h_and_w)

# Return prediction
if p_male_h_and_w > p_female_h_and_w:
  print("class = Male")
else:
  print("class = Female")