Titanic: Machine Learning from Disaster

This example is pandas-ml transration based on kaggle's "Getting Started With Python II". To run this Jupyter Notebook, you must download data to titanic_data directory.

https://www.kaggle.com/c/titanic/details/getting-started-with-python-ii



In [1]:

    
import os

import numpy as np
import pandas as pd

pd.set_option('display.max_rows', 6)



In [2]:

    
import pandas_ml as pdml



In [3]:

    
train_df = pd.read_csv(os.path.join('titanic_data', 'train.csv'), header=0, index_col=0)
test_df = pd.read_csv(os.path.join('titanic_data', 'test.csv'), header=0, index_col=0)

# convert to pdml.ModelFrame
train_df = pdml.ModelFrame(train_df, target='Survived')
test_df = pdml.ModelFrame(test_df)
train_df









    Out[3]:






  
    
      
      Survived
      Pclass
      Name
      Sex
      Age
      SibSp
      Parch
      Ticket
      Fare
      Cabin
      Embarked
    
    
      PassengerId
      
      
      
      
      
      
      
      
      
      
      
    
  
  
    
      1
      0
      3
      Braund, Mr. Owen Harris
      male
      22
      1
      0
      A/5 21171
      7.2500
      NaN
      S
    
    
      2
      1
      1
      Cumings, Mrs. John Bradley (Florence Briggs Th...
      female
      38
      1
      0
      PC 17599
      71.2833
      C85
      C
    
    
      3
      1
      3
      Heikkinen, Miss. Laina
      female
      26
      0
      0
      STON/O2. 3101282
      7.9250
      NaN
      S
    
    
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
    
    
      889
      0
      3
      Johnston, Miss. Catherine Helen "Carrie"
      female
      NaN
      1
      2
      W./C. 6607
      23.4500
      NaN
      S
    
    
      890
      1
      1
      Behr, Mr. Karl Howell
      male
      26
      0
      0
      111369
      30.0000
      C148
      C
    
    
      891
      0
      3
      Dooley, Mr. Patrick
      male
      32
      0
      0
      370376
      7.7500
      NaN
      Q
    
  

891 rows × 11 columns



In [4]:

    
# display target columns
train_df.target









    Out[4]:





PassengerId
1      0
2      1
3      1
      ..
889    0
890    1
891    0
Name: Survived, dtype: int64



In [5]:

    
# display data columns
train_df.data









    Out[5]:






  
    
      
      Pclass
      Name
      Sex
      Age
      SibSp
      Parch
      Ticket
      Fare
      Cabin
      Embarked
    
    
      PassengerId
      
      
      
      
      
      
      
      
      
      
    
  
  
    
      1
      3
      Braund, Mr. Owen Harris
      male
      22
      1
      0
      A/5 21171
      7.2500
      NaN
      S
    
    
      2
      1
      Cumings, Mrs. John Bradley (Florence Briggs Th...
      female
      38
      1
      0
      PC 17599
      71.2833
      C85
      C
    
    
      3
      3
      Heikkinen, Miss. Laina
      female
      26
      0
      0
      STON/O2. 3101282
      7.9250
      NaN
      S
    
    
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
    
    
      889
      3
      Johnston, Miss. Catherine Helen "Carrie"
      female
      NaN
      1
      2
      W./C. 6607
      23.4500
      NaN
      S
    
    
      890
      1
      Behr, Mr. Karl Howell
      male
      26
      0
      0
      111369
      30.0000
      C148
      C
    
    
      891
      3
      Dooley, Mr. Patrick
      male
      32
      0
      0
      370376
      7.7500
      NaN
      Q
    
  

891 rows × 10 columns



In [6]:

    
train_df['Gender'] = train_df['Sex'].replace({'female': 0, 'male': 1})
train_df['Gender']









    Out[6]:





PassengerId
1      1
2      0
3      0
      ..
889    0
890    1
891    1
Name: Gender, dtype: int64



In [7]:

    
if train_df['Embarked'].isnull().any():
    train_df['Embarked'] = train_df['Embarked'].fillna(train_df['Embarked'].value_counts().argmax())
                                                       
ports_factor, ports = train_df['Embarked'].factorize()
ports_dict = dict(zip(ports, np.arange(len(ports))))
ports_dict









    Out[7]:





{'C': 1, 'Q': 2, 'S': 0}



In [8]:

    
train_df['Embarked'] = train_df['Embarked'].replace(ports_dict)



In [9]:

    
if train_df['Age'].isnull().any():
    train_df.loc[ (train_df.Age.isnull()), 'Age'] = train_df['Age'].median()

train_df = train_df.drop(['Name', 'Sex', 'Ticket', 'Cabin'], axis=1) 
train_df









    Out[9]:






  
    
      
      Survived
      Pclass
      Age
      SibSp
      Parch
      Fare
      Embarked
      Gender
    
    
      PassengerId
      
      
      
      
      
      
      
      
    
  
  
    
      1
      0
      3
      22
      1
      0
      7.2500
      0
      1
    
    
      2
      1
      1
      38
      1
      0
      71.2833
      1
      0
    
    
      3
      1
      3
      26
      0
      0
      7.9250
      0
      0
    
    
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
    
    
      889
      0
      3
      28
      1
      2
      23.4500
      0
      0
    
    
      890
      1
      1
      26
      0
      0
      30.0000
      1
      1
    
    
      891
      0
      3
      32
      0
      0
      7.7500
      2
      1
    
  

891 rows × 8 columns



In [10]:

    
test_df['Gender'] = test_df['Sex'].replace({'female': 0, 'male': 1})

if test_df['Embarked'].isnull().any() > 0:
    test_df['Embarked'] = test_df['Embarked'].fillna(test_df['Embarked'].value_counts().argmax())
    
test_df['Embarked'] = test_df['Embarked'].replace(ports_dict)

if test_df['Age'].isnull().any():
    test_df['Age'] = test_df['Age'].fillna(test_df['Age'].median())

if test_df['Fare'].isnull().any():
    median_fare = np.zeros(3)
    for f in range(0,3):                                           
        median_fare[f] = test_df[ test_df.Pclass == f+1 ]['Fare'].median()
    for f in range(0,3):                                        
        test_df.loc[ (test_df.Fare.isnull()) & (test_df.Pclass == f+1 ), 'Fare'] = median_fare[f]

test_df = test_df.drop(['Name', 'Sex', 'Ticket', 'Cabin'], axis=1) 
test_df









    Out[10]:






  
    
      
      Pclass
      Age
      SibSp
      Parch
      Fare
      Embarked
      Gender
    
    
      PassengerId
      
      
      
      
      
      
      
    
  
  
    
      892
      3
      34.5
      0
      0
      7.8292
      2
      1
    
    
      893
      3
      47.0
      1
      0
      7.0000
      0
      0
    
    
      894
      2
      62.0
      0
      0
      9.6875
      2
      1
    
    
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
    
    
      1307
      3
      38.5
      0
      0
      7.2500
      0
      1
    
    
      1308
      3
      27.0
      0
      0
      8.0500
      0
      1
    
    
      1309
      3
      27.0
      1
      1
      22.3583
      1
      1
    
  

418 rows × 7 columns



In [11]:

    
# Training...
forest = train_df.ensemble.RandomForestClassifier(n_estimators=100)
forest = train_df.fit(forest)
forest









    Out[11]:





RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',
            max_depth=None, max_features='auto', max_leaf_nodes=None,
            min_samples_leaf=1, min_samples_split=2,
            min_weight_fraction_leaf=0.0, n_estimators=100, n_jobs=1,
            oob_score=False, random_state=None, verbose=0,
            warm_start=False)



In [12]:

    
# Predicting...
predicted = test_df.predict(forest)
predicted









    Out[12]:





PassengerId
892     0
893     0
894     0
       ..
1307    0
1308    0
1309    1
dtype: int64



In [13]:

    
predicted.reset_index()









    Out[13]:






  
    
      
      PassengerId
      0
    
  
  
    
      0
      892
      0
    
    
      1
      893
      0
    
    
      2
      894
      0
    
    
      ...
      ...
      ...
    
    
      415
      1307
      0
    
    
      416
      1308
      0
    
    
      417
      1309
      1
    
  

418 rows × 2 columns



In [14]:

    
predicted.value_counts()









    Out[14]:





0    261
1    157
dtype: int64

	Survived	Pclass	Name	Sex	Age	SibSp	Parch	Ticket	Fare	Cabin	Embarked
PassengerId
1	0	3	Braund, Mr. Owen Harris	male	22	1	0	A/5 21171	7.2500	NaN	S
2	1	1	Cumings, Mrs. John Bradley (Florence Briggs Th...	female	38	1	0	PC 17599	71.2833	C85	C
3	1	3	Heikkinen, Miss. Laina	female	26	0	0	STON/O2. 3101282	7.9250	NaN	S
...	...	...	...	...	...	...	...	...	...	...	...
889	0	3	Johnston, Miss. Catherine Helen "Carrie"	female	NaN	1	2	W./C. 6607	23.4500	NaN	S
890	1	1	Behr, Mr. Karl Howell	male	26	0	0	111369	30.0000	C148	C
891	0	3	Dooley, Mr. Patrick	male	32	0	0	370376	7.7500	NaN	Q

	Pclass	Age	SibSp	Parch	Fare	Embarked	Gender
PassengerId
892	3	34.5	0	0	7.8292	2	1
893	3	47.0	1	0	7.0000	0	0
894	2	62.0	0	0	9.6875	2	1
...	...	...	...	...	...	...	...
1307	3	38.5	0	0	7.2500	0	1
1308	3	27.0	0	0	8.0500	0	1
1309	3	27.0	1	1	22.3583	1	1