项目目标

泰坦尼克号的沉没是历史上最著名的还难事件之一,在船上的2224名乘客和机组人员中,共造成1502人死亡。本次项目的目标是运用机器学习工具来预测哪些乘客能够幸免于难。

项目过程

  • 导入并探索数据
  • 处理缺失值,删除与预测无关的特征
  • 将分类变量转换为数值型变量
  • 实例化模型并进行交叉验证
  • 模型预测
  • 调参,得到最好的超参数

项目代码(Jupyter)

import pandas as pd
from sklearn.tree import DecisionTreeClassifier
from sklearn.model_selection import train_test_split
from sklearn.model_selection import GridSearchCV
from sklearn.model_selection import cross_val_score
import matplotlib.pyplot as plt
data = pd.read_csv("Taitanic data.csv")
data.head()

【决策树】泰坦尼克号幸存者预测项目-LMLPHP

data.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 891 entries, 0 to 890
Data columns (total 12 columns):
 #   Column       Non-Null Count  Dtype
---  ------       --------------  -----
 0   PassengerId  891 non-null    int64
 1   Survived     891 non-null    int64
 2   Pclass       891 non-null    int64
 3   Name         891 non-null    object
 4   Sex          891 non-null    object
 5   Age          714 non-null    float64
 6   SibSp        891 non-null    int64
 7   Parch        891 non-null    int64
 8   Ticket       891 non-null    object
 9   Fare         891 non-null    float64
 10  Cabin        204 non-null    object
 11  Embarked     889 non-null    object
dtypes: float64(2), int64(5), object(5)
memory usage: 83.7+ KB
# 这里首先我们看看这些标签代表着什么

# PassengerId => 乘客ID
# Pclass => 乘客等级(1/2/3等舱位)
# Name => 乘客姓名
# Sex => 性别
# Age => 年龄
# SibSp => 堂兄弟/妹个数
# Parch => 父母与小孩个数
# Ticket => 船票信息
# Fare => 票价
# Cabin => 客舱
# Embarked => 登船港口
# 删除缺失值过多的列,以及和预测的y没有关系的列
data.drop(["Cabin", "Name", "Ticket"], inplace=True, axis=1)

# 处理缺失值,对于缺失较多的列进行填补,对于缺失较少的列可以直接删除该条记录
data["Age"] = data["Age"].fillna(data["Age"].mean())
data = data.dropna()

# 删除缺失值后重置索引
data.index = range(len(data))
data.tail()

【决策树】泰坦尼克号幸存者预测项目-LMLPHP

data.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 889 entries, 0 to 888
Data columns (total 9 columns):
 #   Column       Non-Null Count  Dtype
---  ------       --------------  -----
 0   PassengerId  889 non-null    int64
 1   Survived     889 non-null    int64
 2   Pclass       889 non-null    int64
 3   Sex          889 non-null    object
 4   Age          889 non-null    float64
 5   SibSp        889 non-null    int64
 6   Parch        889 non-null    int64
 7   Fare         889 non-null    float64
 8   Embarked     889 non-null    object
dtypes: float64(2), int64(5), object(2)
memory usage: 62.6+ KB
# 将分类变量转换为数值型变量

# 在说分类变量转数值型变量之前,我们首先要清楚数据的类型
# 数据可以分为定量数据和定性数据,定性数据又可以分为有序数据和无序数据,定量数据可以分为离散型数据和连续型数据
# 这个项目中我们要处理的数据是Sex和Embarked,前者属于定性数据中的无序数据,后者属于定性数据中的有序数据
# 在sklearn中可以进行变量转换的类有三个:OneHotEncoder\OrdinalEncoder\LableEncoder

# 三者的区别在于:
# 1.OneHotEncoder用于编码无序数据(针对特征)
# 2.OrdinalEncoder用于编码有序数据(针对特征),可以保留数据的大小意义
# 3.LableEncoder用于编码标签变量,不会保留数据的大小意义
#将分类变量转换为数值型变量
from sklearn.preprocessing import OneHotEncoder,OrdinalEncoder

# 编码Sex
ohe = OneHotEncoder(sparse=False)
data_Sex = ohe.fit_transform(data["Sex"].values.reshape(-1, 1))

# 查看编码后对应的特征名称并转换为DataFrame
ohe.get_feature_names()
data_Sex_df = pd.DataFrame(data_Sex, columns=["female","male"])

# 编码Embarked并转换为DataFrame
oe = OrdinalEncoder()
data_Embarked = oe.fit_transform(data["Embarked"].values.reshape(-1, 1))
data_Embarked_df = pd.DataFrame(data_Embarked, columns=["Embarked"])
print(data_Sex.shape)
print(data_Embarked.shape)
print(data.shape)
(889, 2)
(889, 1)
(889, 9)
# 删除Sex和Embarked
data.drop(["Sex", "Embarked"], inplace=True, axis=1)

# 将编码后的数据合并到原数据
newdata = pd.concat([data, data_Sex_df, data_Embarked_df], axis=1)
newdata

【决策树】泰坦尼克号幸存者预测项目-LMLPHP

# 划分特征与标签
X = newdata.iloc[:, newdata.columns != "Survived"]
y = newdata.iloc[:,newdata.columns == "Survived"]

# 划分训练集与测试集
Xtrain, Xtest, Ytrain, Ytest = train_test_split(X,y)

clf = DecisionTreeClassifier(random_state=666)
clf.fit(Xtrain, Ytrain)
score_ = clf.score(Xtest, Ytest)

score_
0.7713004484304933
# 交叉验证
cv_score = []
for i in range(2,10):
    score = cross_val_score(clf,X,y,cv=i).mean()
    cv_score.append(score)

best_cv = cv_score.index(max(cv_score)) + 2
best_cv
5
# 网格搜索
parameters = {"splitter":('best','random')
              ,"max_depth":[*range(1,5)]
              ,"min_samples_leaf":[*range(1,10)]
             }

clf = DecisionTreeClassifier(random_state=666)
GS = GridSearchCV(clf, parameters, cv=best_cv)
GS.fit(Xtrain,Ytrain)

GS.best_score_
0.8138143867130513
GS.best_params_
{'max_depth': 3, 'min_samples_leaf': 1, 'splitter': 'best'}
02-21 22:10