Kaggle Submission: Titanic

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

#Import training data
df_test = pd.read_csv('test.csv')
df_train = pd.read_csv('train.csv')

df_test.head()

df_train.head()

#We have some missing data here, so we're going to map out where NaN exists
#We might be able to take care of age, but cabin is probably too bad to save
sns.heatmap(df_train.isnull(),yticklabels=False,cbar=False,cmap='viridis')

#Test dataset
sns.heatmap(df_test.isnull(),yticklabels=False,cbar=False,cmap='viridis')

#Determine the average ages of passengers by class
#In an attempt to fix the NaN for this column somewhat
sns.set_style('whitegrid')
plt.figure(figsize=(10,7))
sns.boxplot(x='Pclass',y='Age',data=df_train)

#Fill out the Age column with the average ages of passengers per class
def impute_age(cols):
    Age = cols[0]
    Pclass = cols[1]
    
    if pd.isnull(Age):
        if Pclass == 1:
            return 37 #Return the avg age of passengers in the 1st class
        elif Pclass == 2:
            return 29 #2nd class
        else:
            return 24 #3rd class
    else:
        return Age
    
#Apply the function to the Age column
df_train['Age'] = df_train[['Age','Pclass']].apply(impute_age,axis=1)
df_test['Age'] = df_test[['Age','Pclass']].apply(impute_age,axis=1)

#Drop the cabin data
df_train.drop('Cabin',axis=1,inplace=True)
df_test.drop('Cabin',axis=1,inplace=True)

#Recheck the heatmap
#No more problems with Age
sns.heatmap(df_train.isnull(),yticklabels=False,cbar=False,cmap='viridis')

sns.heatmap(df_test.isnull(),yticklabels=False,cbar=False,cmap='viridis')

df_test[df_test['Fare'].isnull() == True]

df_test['Fare'][df_test['Pclass'] == 3].mean()

df_test['Fare'][df_test['PassengerId'] == 1044] = 12.45

#Determine where most people in PClass = 1 embarked
#In an attempt to fix the NaN for this column somewhat
sns.set_style('whitegrid')
plt.figure(figsize=(10,7))
sns.countplot(x="Embarked", data=df_train[df_train['Pclass'] == 1])

df_train[df_train['Embarked'].isnull() == True]

#Determine where most people in PClass = 1 embarked
#In an attempt to fix the NaN for this column somewhat
sns.set_style('whitegrid')
plt.figure(figsize=(10,7))
sns.countplot(x="Embarked", data=df_train[df_train['Pclass'] == 1])

df_train['Embarked'][df_train['Embarked'].isnull() == True] = 'S'

df_train[df_train['Embarked'].isnull() == True]

#Transform male/female into numeric columns - we drop female since M/F are perfect predictors
#Transform embarked into numeric columns Q/S
sex_train = pd.get_dummies(df_train['Sex'],drop_first=True)
embark_train = pd.get_dummies(df_train['Embarked'],drop_first=True)

#Repeat for test dataset
sex_test = pd.get_dummies(df_test['Sex'],drop_first=True)
embark_test = pd.get_dummies(df_test['Embarked'],drop_first=True)

#Add the new columns to the dataset
df_train = pd.concat([df_train,sex_train,embark_train],axis=1)
df_test = pd.concat([df_test,sex_test,embark_test],axis=1)

#Drop the old Sex/Embarked columns, along with the other columns we can't use for predictions
#Save passenger ID's for submission
pass_id = df_test['PassengerId']

df_train.drop(['PassengerId','Sex','Embarked','Name','Ticket'],axis=1,inplace=True)
df_test.drop(['PassengerId','Sex','Embarked','Name','Ticket'],axis=1,inplace=True)

df_train.head()

df_test.head()

#Assign variables
X = df_train.drop('Survived', axis=1)
y = df_train['Survived']

from sklearn.model_selection import train_test_split

#Choose the test size
#Test size = % of dataset allocated for testing (.3 = 30%)
#Random state = # of random splits
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=101)

from sklearn.ensemble import RandomForestClassifier

#Object, try out 200 estimators to start
rfc = RandomForestClassifier(n_estimators=200)

#Fit model
rfc.fit(X_train,y_train)

#Form predictions
predictions = rfc.predict(X_test)

from sklearn.metrics import classification_report, confusion_matrix

#Print reports
print(confusion_matrix(y_test,predictions))
print(classification_report(y_test,predictions))

#Imports
from sklearn.model_selection import RandomizedSearchCV

#Provide a dictionary of these values to test
param_grid = {'bootstrap': [True, False],
 'max_depth': [10, 20, 30, 40, 50, 60, 70, 80, 90, 100, None],
 'max_features': ['auto', 'sqrt'],
 'min_samples_leaf': [1, 2, 4],
 'min_samples_split': [2, 5, 10],
 'n_estimators': [200, 400, 600, 800, 1000, 1200, 1400, 1600, 1800, 2000]}

#Instantiate object
grid = RandomizedSearchCV(estimator = RandomForestClassifier(), param_distributions = param_grid, n_iter = 100, cv = 3, verbose=2, random_state=42, n_jobs = -1)

#Fit to find the best combo of params
grid.fit(X_train,y_train)

#Show the best params to use
grid.best_params_

#New predictions
grid_predictions = grid.predict(X_test)

#Print reports
print(confusion_matrix(y_test,grid_predictions))
print(classification_report(y_test,grid_predictions))

#Object, try out 200 estimators to start
rfc_full = RandomForestClassifier(n_estimators=2000,
                                  min_samples_split=2,
                                  min_samples_leaf=2,
                                  max_features='auto',
                                  max_depth=10,
                                  bootstrap=False)

#Fit model
rfc_full.fit(X,y)

#Run on the test dataframe
full_predictions = rfc_full.predict(df_test)

full_predictions

#Merge into final dataframe
pred_df = pd.DataFrame(full_predictions)

#Drop indexes (can cause NaN when using Concat if you don't do this beforehand)
pred_df.reset_index(drop=True, inplace=True)
pass_id.reset_index(drop=True, inplace=True)

submission = pd.concat([pass_id.iloc[0:],pred_df], axis=1)
submission.columns=['PassengerId','Survived']
submission.to_csv('submission.csv', index = False)

submission.head()