CSI assignment 2

gistfile1.txt

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Sat Dec 5 00:24:39 2021

@author: adwaitnaik
"""

import pandas as pd
import numpy as np
pd.options.mode.chained_assignment = None  # default='warn'

def train_test(X1_train, X2_train, X_test):
    
    #Compute mean and standard deviation for Class == 'smile'
    smile_mean = X1_train.mean(axis=0)
    smile_std = X1_train.std(axis=0)
    
    #Compute mean and standard deviation for Class == 'frown'
    frown_mean = X2_train.mean(axis=0)
    frown_std = X2_train.std(axis=0)
    
    log_P_Xi_C1 = np.zeros(X_test.shape)
    log_P_Xi_C2 = np.zeros(X_test.shape)
    log_P_X_C1 = np.zeros(X_test.shape[0],)
    log_P_X_C2 = np.zeros(X_test.shape[0],)
    
    
    #calculating the likelihood P(Xi|Ck=1)
    for i in range(len(X_test.columns)):
        logA = (np.sqrt(2*np.pi)*smile_std.iloc[i])
        logB = np.square(X_test.iloc[:,i] - smile_mean.iloc[i])/(2*np.square(smile_std.iloc[i]))
        logP = np.log(logA)+logB
        #Log likelihood of observing the 17 features when class is smile
        log_P_Xi_C1[:,i] = logP

    log_P_X_C1 = -np.sum(log_P_Xi_C1,axis=1)
    
    # print(log_P_X_C1)   
    
    #calculating the likelihood P(Xi|Ck=0)
    for i in range(len(X_test.columns)):
        
        logA = (np.sqrt(2*np.pi)*frown_std.iloc[i])
        logB = np.square(X_test.iloc[:,i] - frown_mean.iloc[i])/(2*np.square(frown_std.iloc[i]))
        logP = np.log(logA)+logB
        #Log likelihood of observing the 17 features when class is frown
        log_P_Xi_C2[:,i] = logP
        
    log_P_X_C2 = -np.sum(log_P_Xi_C2,axis=1)
    
    
    #Gaussian Discriminant Function Rule    
    #discriminant function implementation
    gamma_prime_1 = log_P_X_C1 + np.log(Ck)
    gamma_prime_0 = log_P_X_C2 + np.log(Ck) 
    
    G = np.column_stack((gamma_prime_0, gamma_prime_1)) 
    k = np.argmax(G, axis=1)
    
    return k

if __name__ == "__main__":    
    
    #Load data
    test_path = r'/Users/adwaitnaik/test-part-2.csv' #insert path to the test file
    training_path = r'/Users/adwaitnaik/training-part-2.csv' #insert path to the train file'
    
    test_data = pd.read_csv(test_path)
    train_data = pd.read_csv(training_path)
    
    #Separate data according to class     
    smile_train = train_data.loc[train_data["Class"] == 'smile']          
    frown_train = train_data.loc[train_data["Class"] == 'frown']      
    
    #Separate features from class of test data
    X_test = test_data.iloc[:,0:17]
    y_test = test_data.iloc[:,17]
    
    y_test[y_test == 'smile'] = 1
    y_test[y_test == 'frown'] = 0
    
    Ck =  0.5
    
    #Training and testing
    result = train_test(smile_train, frown_train, X_test)
    
    classified_total = len(y_test)
    classified_correct = 0
    classified_incorrect = 0
    
    for i in range(classified_total):
        if y_test[i]==result[i]:
            classified_correct = classified_correct + 1 #smile
        else:
            classified_incorrect = classified_incorrect + 1 #frown
    
    accuracy = classified_correct/classified_total
    error_rate = 1 - accuracy
    
    print("The accuracy is:",accuracy)
    print("The error rate is:",error_rate)