import numpy as np
import pandas as pd
import matplotlib as mpl
import matplotlib.pyplot as plt
import matplotlib.pylab as pylab
import seaborn as sns
%matplotlib inline

#Cria as funções para mostrar os gráficos
#Source https://www.kaggle.com/sachinkulkarni/titanic/an-interactive-data-science-tutorial
def plot_correlation_map( df ):
    corr = titanic.corr()
    _ , ax = plt.subplots( figsize =( 12 , 10 ) )
    cmap = sns.diverging_palette( 220 , 10 , as_cmap = True )
    _ = sns.heatmap(
        corr, 
        cmap = cmap,
        square=True, 
        cbar_kws={ 'shrink' : .9 }, 
        ax=ax, 
        annot = True, 
        annot_kws = { 'fontsize' : 12 }
    )

def plot_line(loss, ylabel, xlabel='Epochs'):
    fig = plt.figure()
    plt.plot(loss)
    plt.ylabel(ylabel)
    plt.xlabel(xlabel)
    plt.show()
    plt.show()

def plot_histograms( df , variables , n_rows , n_cols ):
    fig = plt.figure( figsize = ( 16 , 12 ) )
    for i, var_name in enumerate( variables ):
        ax=fig.add_subplot( n_rows , n_cols , i+1 )
        df[ var_name ].hist( bins=10 , ax=ax )
        ax.set_title( 'Skew: ' + str( round( float( df[ var_name ].skew() ) , ) ) ) # + ' ' + var_name ) #var_name+" Distribution")
        ax.set_xticklabels( [] , visible=False )
        ax.set_yticklabels( [] , visible=False )
    fig.tight_layout()  # Improves appearance a bit.
    plt.show()

def plot_distribution( df , var , target , **kwargs ):
    row = kwargs.get( 'row' , None )
    col = kwargs.get( 'col' , None )
    facet = sns.FacetGrid( df , hue=target , aspect=4 , row = row , col = col )
    facet.map( sns.kdeplot , var , shade= True )
    facet.set( xlim=( 0 , df[ var ].max() ) )
    facet.add_legend()

def plot_categories( df , cat , target , **kwargs ):
    row = kwargs.get( 'row' , None )
    col = kwargs.get( 'col' , None )
    facet = sns.FacetGrid( df , row = row , col = col )
    facet.map( sns.barplot , cat , target )
    facet.add_legend()

train = pd.read_csv("train.csv")
test    = pd.read_csv("test.csv")
full = train.append( test , ignore_index = True )
titanic = full[ :891 ]

titanic.head()

titanic.describe()

#print(titanic.mean())
#print(titanic.std())
#print(titanic.sem())
plot_correlation_map(titanic)

plot_distribution( titanic , var = 'Age' , target = 'Survived' , row = 'Sex' )

plot_categories( titanic , cat = 'Embarked' , target = 'Survived' )

plot_categories( titanic , cat = 'Sex' , target = 'Survived' )

plot_categories( titanic , cat = 'Pclass' , target = 'Survived' )

plot_categories( titanic , cat = 'SibSp' , target = 'Survived' )

plot_categories( titanic , cat = 'Parch' , target = 'Survived' )

sex = pd.Series( np.where( full.Sex == 'male' , 1 , 0 ) , name = 'Sex' )

embarked = pd.get_dummies( full.Embarked , prefix='Embarked' )
embarked.head()

pclass = pd.get_dummies( full.Pclass , prefix='Pclass' )
pclass.head()

imputed = pd.DataFrame()
imputed[ 'Age' ] = full.Age.fillna( full.Age.mean() )
imputed[ 'Fare' ] = full.Fare.fillna( full.Fare.mean() )
#imputed['SibSp'] = full.SibSp.fillna( full.SibSp.mean() )
#imputed['Parch'] = full.Parch.fillna( full.Parch.mean() )
imputed.head()

cabin = pd.DataFrame()

#Substitui os dados de cabines faltantes com U (Unknown)
cabin[ 'Cabin' ] = full.Cabin.fillna( 'U' )

# mapping each Cabin value with the cabin letter
cabin[ 'Cabin' ] = cabin[ 'Cabin' ].map( lambda c : c[0] )
cabin = pd.get_dummies( cabin['Cabin'] , prefix = 'Cabin' )

cabin.head()

full_X = pd.concat( [ imputed , embarked , pclass , sex, cabin ] , axis=1 )
full_X.tail()

from sklearn.model_selection import train_test_split

full_X_norm = full_X.apply(lambda x: (x - np.mean(x)) / (np.max(x) - np.min(x)))
full_X['Age'] = full_X_norm['Age']
full_X['Fare'] = full_X_norm['Fare']
#full_X['SibSp'] = full_X_norm['SibSp']
#full_X['Parch'] = full_X_norm['Parch']

# Create all datasets that are necessary to train, validate and test models
train_valid_X = full_X[ 0:891 ]
train_valid_y = titanic.Survived
test_X = full_X[ 891: ]
train_X , valid_X , train_y , valid_y = train_test_split( train_valid_X , train_valid_y , train_size = .7 )

import warnings
warnings.filterwarnings('ignore')
from __future__ import print_function
from sklearn.svm import SVC, LinearSVC
from sklearn.metrics import classification_report
from sklearn.model_selection import GridSearchCV

tuned_parameters = [{'kernel': ['rbf'], 'gamma': [1e-2, 1e-3, 1e-4, 1e-5],
                     'C': [1, 10, 100, 1000]},
                    {'kernel': ['linear'], 'C': [1, 10, 100, 1000]}]

scores = ['precision', 'recall']

for score in scores:
    model = GridSearchCV(SVC(C=1), tuned_parameters, cv=5,
                           scoring='%s_macro' % score)
    model.fit(train_X , train_y)

    # Score the model
    print (model.score( train_X , train_y ) , model.score( valid_X , valid_y ))

    print("Best parameters set found on development set:")
    print()
    print(model.best_params_)
    print()
    print("Grid scores on development set:")
    print()
    means = model.cv_results_['mean_test_score']
    stds = model.cv_results_['std_test_score']
    for mean, std, params in zip(means, stds, model.cv_results_['params']):
        print("%0.3f (+/-%0.03f) for %r"
              % (mean, std * 2, params))
    print()

    print("Detailed classification report:")
    print()
    print("The model is trained on the full development set.")
    print("The scores are computed on the full evaluation set.")
    print()
    y_true, y_pred = valid_y, model.predict(valid_X)
    print(classification_report(y_true, y_pred))
    print()

0.8378125 0.841176470588
Best parameters set found on development set:

{'C': 1000, 'gamma': 0.01, 'kernel': 'rbf'}

Grid scores on development set:

0.778 (+/-0.073) for {'C': 1, 'gamma': 0.01, 'kernel': 'rbf'}
0.306 (+/-0.002) for {'C': 1, 'gamma': 0.001, 'kernel': 'rbf'}
0.306 (+/-0.002) for {'C': 1, 'gamma': 0.0001, 'kernel': 'rbf'}
0.306 (+/-0.002) for {'C': 1, 'gamma': 1e-05, 'kernel': 'rbf'}
0.778 (+/-0.073) for {'C': 10, 'gamma': 0.01, 'kernel': 'rbf'}
0.778 (+/-0.073) for {'C': 10, 'gamma': 0.001, 'kernel': 'rbf'}
0.306 (+/-0.002) for {'C': 10, 'gamma': 0.0001, 'kernel': 'rbf'}
0.306 (+/-0.002) for {'C': 10, 'gamma': 1e-05, 'kernel': 'rbf'}
0.770 (+/-0.081) for {'C': 100, 'gamma': 0.01, 'kernel': 'rbf'}
0.778 (+/-0.073) for {'C': 100, 'gamma': 0.001, 'kernel': 'rbf'}
0.778 (+/-0.073) for {'C': 100, 'gamma': 0.0001, 'kernel': 'rbf'}
0.306 (+/-0.002) for {'C': 100, 'gamma': 1e-05, 'kernel': 'rbf'}
0.818 (+/-0.067) for {'C': 1000, 'gamma': 0.01, 'kernel': 'rbf'}
0.770 (+/-0.081) for {'C': 1000, 'gamma': 0.001, 'kernel': 'rbf'}
0.778 (+/-0.073) for {'C': 1000, 'gamma': 0.0001, 'kernel': 'rbf'}
0.778 (+/-0.073) for {'C': 1000, 'gamma': 1e-05, 'kernel': 'rbf'}
0.770 (+/-0.081) for {'C': 1, 'kernel': 'linear'}
0.770 (+/-0.081) for {'C': 10, 'kernel': 'linear'}
0.770 (+/-0.081) for {'C': 100, 'kernel': 'linear'}
0.770 (+/-0.081) for {'C': 1000, 'kernel': 'linear'}

Detailed classification report:

The model is trained on the full development set.
The scores are computed on the full evaluation set.

             precision    recall  f1-score   support

        0.0       0.80      0.95      0.87       168
        1.0       0.88      0.60      0.71       100

avg / total       0.83      0.82      0.81       268


0.774956074706 0.785714285714
Best parameters set found on development set:

{'C': 1, 'kernel': 'linear'}

Grid scores on development set:

0.762 (+/-0.072) for {'C': 1, 'gamma': 0.01, 'kernel': 'rbf'}
0.500 (+/-0.000) for {'C': 1, 'gamma': 0.001, 'kernel': 'rbf'}
0.500 (+/-0.000) for {'C': 1, 'gamma': 0.0001, 'kernel': 'rbf'}
0.500 (+/-0.000) for {'C': 1, 'gamma': 1e-05, 'kernel': 'rbf'}
0.762 (+/-0.072) for {'C': 10, 'gamma': 0.01, 'kernel': 'rbf'}
0.762 (+/-0.072) for {'C': 10, 'gamma': 0.001, 'kernel': 'rbf'}
0.500 (+/-0.000) for {'C': 10, 'gamma': 0.0001, 'kernel': 'rbf'}
0.500 (+/-0.000) for {'C': 10, 'gamma': 1e-05, 'kernel': 'rbf'}
0.762 (+/-0.079) for {'C': 100, 'gamma': 0.01, 'kernel': 'rbf'}
0.762 (+/-0.072) for {'C': 100, 'gamma': 0.001, 'kernel': 'rbf'}
0.762 (+/-0.072) for {'C': 100, 'gamma': 0.0001, 'kernel': 'rbf'}
0.500 (+/-0.000) for {'C': 100, 'gamma': 1e-05, 'kernel': 'rbf'}
0.761 (+/-0.053) for {'C': 1000, 'gamma': 0.01, 'kernel': 'rbf'}
0.762 (+/-0.079) for {'C': 1000, 'gamma': 0.001, 'kernel': 'rbf'}
0.762 (+/-0.072) for {'C': 1000, 'gamma': 0.0001, 'kernel': 'rbf'}
0.762 (+/-0.072) for {'C': 1000, 'gamma': 1e-05, 'kernel': 'rbf'}
0.763 (+/-0.078) for {'C': 1, 'kernel': 'linear'}
0.762 (+/-0.079) for {'C': 10, 'kernel': 'linear'}
0.762 (+/-0.079) for {'C': 100, 'kernel': 'linear'}
0.762 (+/-0.079) for {'C': 1000, 'kernel': 'linear'}

Detailed classification report:

The model is trained on the full development set.
The scores are computed on the full evaluation set.

             precision    recall  f1-score   support

        0.0       0.85      0.82      0.83       168
        1.0       0.71      0.75      0.73       100

avg / total       0.80      0.79      0.80       268

from sklearn.neural_network import MLPClassifier

model = MLPClassifier(solver='sgd', alpha=1e-5, hidden_layer_sizes=(6, 2), random_state=1)
model.fit( train_X , train_y )

# Score the model
print ("Train: ", model.score( train_X , train_y ))
print ("Test:", model.score( valid_X , valid_y ))

Train:  0.781701444623
Test: 0.798507462687

from sklearn import tree
model = tree.DecisionTreeClassifier()
model.fit(train_X , train_y)
print ("Train: ", model.score( train_X , train_y ))
print ("Test:", model.score( valid_X , valid_y ))

Train:  0.987158908507
Test: 0.813432835821

### Adaboost

from sklearn.ensemble import AdaBoostClassifier
model = AdaBoostClassifier(n_estimators=50)
model.fit(train_X , train_y)
print ("Train: ", model.score( train_X , train_y ))
print ("Test:", model.score( valid_X , valid_y ))

Train:  0.84911717496
Test: 0.832089552239

### Gradient Boosting

from sklearn.ensemble import GradientBoostingClassifier
model = GradientBoostingClassifier(n_estimators=100, learning_rate=1.0, max_depth=1, random_state=0)
model.fit(train_X , train_y)
print ("Train: ", model.score( train_X , train_y ))
print ("Test:", model.score( valid_X , valid_y ))

Train:  0.873194221509
Test: 0.820895522388

from sklearn.naive_bayes import GaussianNB
model = GaussianNB()
model.fit(train_X , train_y)
print ("Train: ", model.score( train_X , train_y ))
print ("Test:", model.score( valid_X , valid_y ))

Train:  0.711075441413
Test: 0.716417910448

from keras.utils.np_utils import to_categorical

t_X = train_X.as_matrix()
t_y = train_y.as_matrix()
#t_y = to_categorical(t_y)

v_X = valid_X.as_matrix()
v_y = valid_y.as_matrix()
#v_y = to_categorical(v_y)

test_X_m = test_X.as_matrix()
print(test_X_m)

Using TensorFlow backend.


[[ 0.05785873 -0.04970687  0.         ...,  0.          0.          1.        ]
 [ 0.21444147 -0.05132536  0.         ...,  0.          0.          1.        ]
 [ 0.40234075 -0.04607971  0.         ...,  0.          0.          1.        ]
 ..., 
 [ 0.10796521 -0.05083739  0.         ...,  0.          0.          1.        ]
 [ 0.         -0.04927589  0.         ...,  0.          0.          1.        ]
 [ 0.         -0.02134795  1.         ...,  0.          0.          1.        ]]

from keras.models import Sequential
from keras.layers import Dense, Dropout, Activation
from keras.optimizers import SGD

model = Sequential()
model.add(Dense(20, input_dim=t_X.shape[1], activation='sigmoid'))
model.add(Dropout(0.5))
model.add(Dense(1, activation='sigmoid'))
model.compile(loss='mse', optimizer='rmsprop', metrics=['accuracy'])

print(model.summary())

____________________________________________________________________________________________________
Layer (type)                     Output Shape          Param #     Connected to                     
====================================================================================================
dense_1 (Dense)                  (None, 20)            380         dense_input_1[0][0]              
____________________________________________________________________________________________________
dropout_1 (Dropout)              (None, 20)            0           dense_1[0][0]                    
____________________________________________________________________________________________________
dense_2 (Dense)                  (None, 1)             21          dropout_1[0][0]                  
====================================================================================================
Total params: 401
Trainable params: 401
Non-trainable params: 0
____________________________________________________________________________________________________
None

hist = model.fit(t_X, t_y, 
          batch_size=64, 
          nb_epoch=200, 
          verbose=False)

plot_line(hist.history['acc'], 'accuracy')

plot_line(hist.history['loss'], 'loss')

loss, accuracy = model.evaluate(v_X, v_y)

print("Loss: ", loss)
print("Accuracy: ", accuracy)

 32/268 [==>...........................] - ETA: 0sLoss:  0.13912824975
Accuracy:  0.828358208955

t_X = test_X.as_matrix()

test_Y = model.predict( t_X )
test_Y = test_Y.flatten().round().astype(int)
passenger_id = full[891:].PassengerId
test = pd.DataFrame( { 'PassengerId': passenger_id , 'Survived': test_Y } )
test.shape
test.head()
test.to_csv( 'titanic_pred.csv' , index = False )