file1

# -*- coding: utf-8 -*-
"""
Created on Thu Oct  6 21:28:37 2016

@author: ben
"""
import numpy

my_dict = {
    'ben': [100, 90, 80, 85],
    'john': [70, 65, 42],
    'mary': [100, 98, 95]
}

names = ['b', 'j', 'm']
scores = [[1,2,3], [4,5,6], [7,8,9]]

for student in my_dict.keys():
    scores = my_dict[student]
    print student, numpy.mean(scores), numpy.std(scores)
    

file2

# -*- coding: utf-8 -*-
"""
Created on Thu Oct  6 21:02:26 2016

@author: ben
"""

#guessing game
#have hidden number
# ask user for guess
#tell use if guess is too high or too low
#
#import random
#
#num_to_guess = random.randint(1,100)
#
#user_input = -1
#counter = 0
#
#while (user_input != num_to_guess):
#    user_input = float(raw_input('Enter your guess: '))
#    counter += 1
#    #counter = counter + 1
#    if user_input > num_to_guess:
#        print("too high!")
#    elif user_input < num_to_guess:
#        print "too low!"
#    else:
#        print "you got it"
#    
#    
##print "This took you " + str(counter) + " tries"
#
#print "This took you {} tries to guess the number {}.".format(counter, num_to_guess)
## same hidden num every time
#

#def myfunc():
#    print 'hello!'
#    x = 5 + 9
#    print x
#    
#def crazy_math(first_number, num_to_multiply):
#    answer = (first_number * 10) - (4 * num_to_multiply)
#    return answer
#    
#print crazy_math(10, 4)
import csv
import numpy

import matplotlib.pyplot as plt

myfile = open('/Users/Ben/Desktop/train.csv')
csv_file = csv.DictReader(myfile)
predictions = []
truths = []

def predict(datapoint):
    # take in a line from our file
    # spit out a single prediction for that line
    print datapoint['Name']
    age = datapoint['Age']
    sex = datapoint['Sex']
    pclass = int(datapoint['Pclass'])
    if age == '':
        return True
    else:
        age = float(age)

        if age <= 15:
            return True
        #if over 15, we need to decide some other way
        else:
            if sex == 'female':
                return True
            else:
                return False
#                if pclass == 1:
#                    return True
#                else:
#                    return False
            

yes = []
no = []

for line in csv_file:
    age = line['Age']
    if line['Survived'] == '1':
        truths.append(True)
    else:
        truths.append(False)
    if age == '':
        pass
    else:
        pclass = line['Pclass']
        if line['Survived'] == '1':
            yes.append(int(pclass))
        else:
            no.append(int(pclass))
            
    mypred = predict(line)
    predictions.append(mypred)
    print '-----'
    
correct = 0.0
l = len(predictions)
for x in range(l):
    if predictions[x] == truths[x]:
        correct += 1

print correct/len(predictions), ' correct answers'

print numpy.mean(yes)
print numpy.mean(no)

yes_hist = numpy.histogram(yes)
no_hist = numpy.histogram(no)

for x in range(len(no_hist)):
    print yes_hist[x] - no_hist[x]

print 'YES'
plt.hist(yes, bins=3)
plt.show()
print 'NO'
plt.hist(no, bins=3)