file.py

#basic guessing game
import csv
import random
import re
import numpy

import matplotlib.pyplot as plt

#num_to_guess = random.randint(0,100)
#
##print "Youre trying to guess {}".format(num_to_guess)
##3 options
##1 - guess = num_to_guess, user wins, guess < num_to_guess, print too little, guess > num_to_guess, print too high
#num_guesses = 0
#guessed = False
#while guessed == False:
#    num_guesses += 1
#    guess = int(raw_input('Enter your number: '))
#
#    if (guess == num_to_guess) or (5 < 7) or ____:
#        print "You got it"
#        print "It took you {num_guesses} guesses to guess the number {num_to_guess}.".format(num_guesses=num_guesses, num_to_guess=num_to_guess)
#        guessed = True
#    
#    elif guess > num_to_guess * 2:
#        print "You guessed WAY too high"
#     
#    elif guess > num_to_guess:
#        print "You guessed too high"
#        
#    else:
#        print "You guessed too low"
#
#
#print "You won!"
#elif guess <= num_to_guess:
#    print "You guessed too low"
#
#ages = []
#myfile = open('/Users/Ben/Desktop/train.csv')
#for line in csv.DictReader(myfile):
#    if line['Age'] != '':
#        ages.append(float(line['Age']))
#
#average = numpy.mean(ages)


classes = []
sexes = []
survived = []
myfile = open('/Users/Ben/Desktop/train.csv')
for line in csv.DictReader(myfile):
    if line['Pclass'] != '' and line['Sex'] != '':
        if line['Survived'] == '1':
            classes.append(float(line['Pclass']))
            if line['Sex'] == 'male':
                sexes.append(1)
            else:
                sexes.append(0)
            survived.append('g')
        else:
            classes.append(float(line['Pclass']))
            if line['Sex'] == 'male':
                sexes.append(1)
            else:
                sexes.append(0)
            survived.append('r')

plt.scatter(classes, sexes, color=survived)

def predict(datapoint):
    #predict takes a datapoint and predicts whether the person will survive
    #first idea: > 18, no < 18, yes
    if line['Age'] != '':
        age = float(datapoint['Age'])
    else:
        age = average
    if line['Sex'] != '':
        if line['Sex'] == 'male':
            male = True
        else:
            male = False
    #2 vars, age, is the age, male is a bool if the passenger was male
    if line['Pclass'] == '1':
        return True
    if age < 18:
        if male == True:
            return True
        elif male == False:
            return False
    elif age > 65:
        return False
        
    else:
        if male == True:
            return False
        else:
            return True
    
    
results = []
myfile = open('/Users/Ben/Desktop/train.csv')
for line in csv.DictReader(myfile):
    print line
    prediction = predict(line)
    #prediction is True or False
    survived = line['Survived']
    #survived is a zero or a one
    if survived == '0':
        survived_bool = False
    else:
        survived_bool = True

    results.append(prediction == survived_bool)
    print '-----'
    
corrects =  results.count(True)
wrongs =  results.count(False)

print corrects/float(corrects + wrongs)