Day 10_DecisionTree_With_Preprocessing

# -*- coding: utf-8 -*-
"""Day 11_decision_tree.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1ybNWDQmSG1mzPaSpZOk8uTdZejQKcRAu
"""
 
from google.colab import drive
drive.mount('/gdrive')
 
PATH = "/gdrive/My Drive/Colab Notebooks/resources/"
 
# %matplotlib inline
 
import numpy as np
import matplotlib.pyplot as plt
import matplotlib as mpl
import pandas as pd
import seaborn as sns
import time
 
 
from scipy.stats import norm, multivariate_normal
# installing packages for interactive graphs
import ipywidgets as widgets
from IPython.display import display
from ipywidgets import interact, interactive, fixed, interact_manual, IntSlider
 
##############################################
################ Helper Function##############
##############################################
 
def my_df_dropNas(df, columns):
    for col in columns:
        df =  df[ df[col].notna() ]
        
def my_checkNas( x ):
    y = x.apply(lambda x : (
    sum(x.isna())
))
    return y
 
@interact_manual( x =IntSlider(0,0,12) )
def test_model( x ):
    print(x)
 
train = pd.read_csv(PATH + "titanic/train.csv")
test  = pd.read_csv(PATH + "titanic/test.csv")
 
##############################################
####### 시리즈 안에 있는 문자열을 맵핑########
##############################################
 
mapping = { "Mr":0 , "Miss":1, "Mrs":2 }
data['title'].map(mapping)
 
train = pd.read_csv(PATH + "titanic/train.csv")
test = pd.read_csv(PATH + "titanic/test.csv")
 
train.describe()
my_checkNas(train)
 
datas = [train, test]
## 파생변수를 만들때는 테스트할때도 사용될 수 있으므로 테스트데이터 에도 똑같이 만들어준다.
 
for data in datas:
    data['Age_mean'] = data['Age']
    data['Age_mean'] = data['Age_mean'].fillna(data['Age_mean'].mean())
 
train['Sex'].value_counts()
 
for data in datas:
    data['Gender'] = data['Sex'] == 'female'
 
my_checkNas(test)
 
## 원핫 인코딩 대체
for data in datas:
    data['Embarked_S'] = data['Embarked'] == "S"
    data['Embarked_C'] = data['Embarked'] == "C"
    data['Embarked_Q'] = data['Embarked'] == "Q"
 
## 가족의 수 컬럼 ( 자기 포함 )
for data in datas:
    data['FamilySize'] = data['SibSp'] + data['Parch'] + 1
 
for data in datas:
    print(data['FamilySize'].value_counts())
    data['Family'] = train['FamilySize']
 
for data in datas:
#대가족(5명이상) : l , 소가족(1명) : S , 중가족( 2 ~ 4 ) : M
    data.loc[  data['FamilySize'] == 1, 'Family' ] = "S"
    data.loc[ (data['FamilySize'] > 1) & (data['FamilySize'] < 5 ), 'Family' ] = "M"
    data.loc[  data['FamilySize'] > 4, 'Family' ] = "L"
 
for data in datas:
    data['Family_S'] = data['Family'] == 'S'
    data['Family_M'] = data['Family'] == 'M'
    data['Family_L'] = data['Family'] == 'L'
 
for data in datas:
    data['Pclass'] = data['Pclass'].astype( 'category' )
 
fn = [ 'Gender','Age_mean','Embarked_S','Embarked_C','Embarked_Q','Family_S','Family_M','Family_L']
 
X_train = train[fn]
Y_train = train['Survived']
 
X_test  = test[fn]
 
# 모델링 ( decision tree )
from sklearn.tree import DecisionTreeClassifier
 
model = DecisionTreeClassifier(max_depth = 3)
 
model.fit(X_train, Y_train)
 
from sklearn.model_selection import KFold, cross_val_score
 
## 셔플 할 때에 random_state => randomseed
kfold = KFold( n_splits = 10, shuffle = True , random_state = 2019)
kfold
 
## cv => cross-validation generator
score = cross_val_score( model, X_train, Y_train, cv = kfold, scoring = 'accuracy' )
score.mean()
 
@interact( x =IntSlider(0,3,100) )
def test(x):
 
    model = DecisionTreeClassifier(max_depth = x, )
    
    model.fit(X_train, Y_train)
    
    ## 셔플 할 때에 random_state => randomseed
    kfold = KFold( n_splits = 100, shuffle = True , random_state = 2019)
    score = cross_val_score( model, X_train, Y_train, cv = kfold, scoring = 'accuracy' )
    print(score.mean())
 
result = model.predict( X_test )
 
test['Survived'] = result.copy()
 
submit = test[['PassengerId','Survived']]
submit.to_csv(PATH + "titanic/submit.csv", index = False )