naive bayes

data.csv

main.py

#!/usr/bin/env python3

import csv
import sys

def main():
    if len(sys.argv) < 2:
        raise ValueError('Requires path to *.csv file')

    name = sys.argv[1]

    data = [] # the hole csv data
    with open(name, 'r') as fd:
        r = csv.reader(fd)
        data = [d for d in r]

    print("Attributes = {}".format(data[0][:-1]))
    print("Classifier = {}".format(data[0][-1]))
    instance_to_classify = ['0', '0']
    y = naive_bayes(data[1:], instance_to_classify)
    print("Instance to classify = {}".format(instance_to_classify))
    print("Bayes naive decision for y =  {}".format(y))
    y = joint_bayes(data[1:], instance_to_classify)
    print("Joint bayes decision for y =  {}".format(y))

def column(data, column):
    return [row[column] for row in data]

def naive_bayes(data, instance):
    target_names  = list(set(column(data, -1)))
    mle = {}
    for target_name in target_names:
        targets = column(data, -1).count(target_name)
        maximum_likelihood = 0
        p = 1
        for index, element in enumerate(instance):
            nominator = 0
            for row in data:
                if (target_name == row[-1] and 
                        element == row[index]):
                    nominator += 1
            p = (p * (nominator / targets))
        maximum_likelihood += (p * (targets/len(data)))
        mle[target_name] = maximum_likelihood

    return max(mle, key=mle.get)

def compare(one, two):
    if len(one) != len(two):
        raise ValueError("Lists needs to have the same length")

    for i, j in zip(one, two):
        if i != j:
            return False

    return True

def joint_bayes(data, instance):
    target_names = list(set(column(data, -1)))
    mle = {}
    for target_name in target_names:
        targets = column(data, -1).count(target_name)
        maximum_likelihood = 0
        nominator = 0
        for row in data:
            if (target_name == row[-1] and 
                compare(instance, row[:-1])):
                    nominator += 1

        mle[target_name] = ((nominator/targets) * (targets/len(data)))
    
    return max(mle, key=mle.get)

if __name__ == '__main__':
    main()