In [1]:

    

import numpy as np
import pandas as pd
from pandas import Series, DataFrame
import sklearn as sk
import matplotlib.pyplot as plt
%matplotlib inline
from __future__ import division
from collections import Counter


    

In [2]:

    

# Read the training data
raw = pd.read_table( '../input/train.csv', delimiter=',')


    

In [3]:

    

raw.head()


    

    
Out[3]:






  

    

      

      
PassengerId
      
Survived
      
Pclass
      
Name
      
Sex
      
Age
      
SibSp
      
Parch
      
Ticket
      
Fare
      
Cabin
      
Embarked
    
  
  

    

      
0
      
 1
      
 0
      
 3
      
                           Braund, Mr. Owen Harris
      
   male
      
 22
      
 1
      
 0
      
        A/5 21171
      
  7.2500
      
  NaN
      
 S
    
    

      
1
      
 2
      
 1
      
 1
      
 Cumings, Mrs. John Bradley (Florence Briggs Th...
      
 female
      
 38
      
 1
      
 0
      
         PC 17599
      
 71.2833
      
  C85
      
 C
    
    

      
2
      
 3
      
 1
      
 3
      
                            Heikkinen, Miss. Laina
      
 female
      
 26
      
 0
      
 0
      
 STON/O2. 3101282
      
  7.9250
      
  NaN
      
 S
    
    

      
3
      
 4
      
 1
      
 1
      
      Futrelle, Mrs. Jacques Heath (Lily May Peel)
      
 female
      
 35
      
 1
      
 0
      
           113803
      
 53.1000
      
 C123
      
 S
    
    

      
4
      
 5
      
 0
      
 3
      
                          Allen, Mr. William Henry
      
   male
      
 35
      
 0
      
 0
      
           373450
      
  8.0500
      
  NaN
      
 S
    
  

In [4]:

    

# Check which columns have missing values
raw.apply(lambda x: sum(x.isnull())/len(x))


    

    
Out[4]:





PassengerId    0.000000
Survived       0.000000
Pclass         0.000000
Name           0.000000
Sex            0.000000
Age            0.198653
SibSp          0.000000
Parch          0.000000
Ticket         0.000000
Fare           0.000000
Cabin          0.771044
Embarked       0.002245
dtype: float64

In [5]:

    

raw.info()


    

    




<class 'pandas.core.frame.DataFrame'>
Int64Index: 891 entries, 0 to 890
Data columns (total 12 columns):
PassengerId    891 non-null int64
Survived       891 non-null int64
Pclass         891 non-null int64
Name           891 non-null object
Sex            891 non-null object
Age            714 non-null float64
SibSp          891 non-null int64
Parch          891 non-null int64
Ticket         891 non-null object
Fare           891 non-null float64
Cabin          204 non-null object
Embarked       889 non-null object
dtypes: float64(2), int64(5), object(5)
memory usage: 90.5+ KB

In [6]:

    

# Look at estimating the missing ages 
ageData = raw.groupby([x.split()[1] for x in raw.Name]).Age
# raw[raw.Age.isnull()]


    

In [7]:

    

def print_cats(df):
    for col in ['Pclass', 'Sex', 'Embarked']:
        print col
        print Counter(df[col]).most_common()


    

In [8]:

    

print_cats(raw)


    

    




Pclass
[(3, 491), (1, 216), (2, 184)]
Sex
[('male', 577), ('female', 314)]
Embarked
[('S', 644), ('C', 168), ('Q', 77), (nan, 2)]

In [9]:

    

def clean(raw):
    cleaned = raw.copy()
    cleaned.loc[cleaned['Age'].isnull(), ['Age']] = np.nanmedian(raw['Age'])
    cleaned.loc[cleaned['Embarked'].isnull(),['Embarked']] = Counter(raw['Embarked']).most_common(1)[0][0]
    # Take group median for missing Fares
    data = cleaned.groupby('Pclass')['Fare']
    cleaned['Fare'] = data.transform( lambda x: x.fillna(x.median()))    
    cleaned['IsChild'] = 1.0*(cleaned['Age'] < 20)
    cleaned['IsFemale'] = 1.0*(cleaned['Sex'] == 'male')
    cleaned['IsUpperClass'] = 1.0*(cleaned['Pclass']==1)
    cleaned['HighFareBucket'] = 1.0*(cleaned['Fare'] >= np.percentile(cleaned['Fare'], 50))
    return cleaned


    

In [10]:

    

cleaned = clean(raw)
cleaned.apply(lambda x: sum(x.isnull())/len(x))


    

    
Out[10]:





PassengerId       0.000000
Survived          0.000000
Pclass            0.000000
Name              0.000000
Sex               0.000000
Age               0.000000
SibSp             0.000000
Parch             0.000000
Ticket            0.000000
Fare              0.000000
Cabin             0.771044
Embarked          0.000000
IsChild           0.000000
IsFemale          0.000000
IsUpperClass      0.000000
HighFareBucket    0.000000
dtype: float64

In [14]:

    

# Histogram of survival vs age
plt.hist([cleaned['Age'][cleaned['Survived']==0],cleaned['Age'][cleaned['Survived']==1]])


    

    
Out[14]:





([array([  18.,   27.,  114.,  229.,   66.,   46.,   24.,   15.,    9.,    1.]),
  array([  36.,   19.,   63.,  117.,   52.,   24.,   21.,    9.,    0.,    1.])],
 array([  0.42 ,   8.378,  16.336,  24.294,  32.252,  40.21 ,  48.168,
         56.126,  64.084,  72.042,  80.   ]),
 <a list of 2 Lists of Patches objects>)






    

In [15]:

    

print_cats(cleaned)


    

    




Pclass
[(3, 491), (1, 216), (2, 184)]
Sex
[('male', 577), ('female', 314)]
Embarked
[('S', 646), ('C', 168), ('Q', 77)]

In [16]:

    

from sklearn.naive_bayes import MultinomialNB
from sklearn.metrics import confusion_matrix, accuracy_score
from sklearn.ensemble import RandomForestClassifier


    

In [17]:

    

nTrain = 891
features = ['IsFemale', 'IsChild', 'Pclass', 'Fare']


    

In [18]:

    

cleaned.groupby(features)['Survived'].mean()


    

    
Out[18]:





IsFemale  IsChild  Pclass  Fare   
0         0        1       25.9292    1
                           26.5500    1
                           27.7208    1
                           28.7125    0
                           31.0000    1
                           39.6000    1
                           49.5000    1
                           49.5042    1
                           51.4792    1
                           51.8625    1
                           52.0000    1
                           52.5542    1
                           53.1000    1
                           55.0000    1
                           55.9000    1
...
1         1        3       15.9000    1.0
                           18.0000    0.0
                           18.7875    0.0
                           20.2125    0.0
                           20.2500    0.0
                           20.5250    1.0
                           20.5750    1.0
                           21.0750    0.0
                           27.9000    0.0
                           29.1250    0.0
                           31.2750    0.0
                           31.3875    0.5
                           34.3750    0.0
                           39.6875    0.0
                           46.9000    0.0
Name: Survived, Length: 430, dtype: float64

In [19]:

    

# clf = MultinomialNB()
clf = RandomForestClassifier()
clf.fit(cleaned[features].values, cleaned['Survived'].values)


    

    
Out[19]:





RandomForestClassifier(bootstrap=True, compute_importances=None,
            criterion='gini', max_depth=None, max_features='auto',
            max_leaf_nodes=None, min_density=None, min_samples_leaf=1,
            min_samples_split=2, n_estimators=10, n_jobs=1,
            oob_score=False, random_state=None, verbose=0)

In [20]:

    

confusion_matrix(cleaned.Survived.values,clf.predict(cleaned[features].values))


    

    
Out[20]:





array([[517,  32],
       [ 48, 294]])

In [21]:

    

accuracy_score(cleaned.Survived.values,clf.predict(cleaned[features].values), normalize=True)


    

    
Out[21]:





0.91021324354657684

In [22]:

    

# Predict the test set
test = pd.read_table('../input/test.csv', delimiter=',')
cleaned_test = clean(test)
cleaned_test['Survived'] = clf.predict(cleaned_test[features].values )


    

In [23]:

    

cleaned_test[['PassengerId', 'Survived']].describe()


    

    
Out[23]:






  

    

      

      
PassengerId
      
Survived
    
  
  

    

      
count
      
  418.000000
      
 418.000000
    
    

      
mean
      
 1100.500000
      
   0.351675
    
    

      
std
      
  120.810458
      
   0.478065
    
    

      
min
      
  892.000000
      
   0.000000
    
    

      
25%
      
  996.250000
      
   0.000000
    
    

      
50%
      
 1100.500000
      
   0.000000
    
    

      
75%
      
 1204.750000
      
   1.000000
    
    

      
max
      
 1309.000000
      
   1.000000
    
  

In [24]:

    

cleaned_test.to_csv('random_forest_prediction_with_pclass.csv', columns=['PassengerId', 'Survived'], index=False)


    

In [ ]: