泰坦尼克号生存预测

一、数据处理

1、数据源包含train数据集和test数据集

2、train数据集包含891条乘客信息，test数据集包含418条乘客信息

3、数据包含以下信息：PassengerId, Survived, Pclass, Name, Sex, Age, SibSp, Parch, Ticket, Fare, Cabin和Embarked

4、数据预处理包括：缺失值填充、数据清洗、特征选择

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

# 读取数据
train_data = pd.read_csv('train.csv')
test_data = pd.read_csv('test.csv')

# 数据预处理
def data_preprocessing(df):
    # 缺失值填充
    df['Age'] = df['Age'].fillna(df['Age'].mean())
    df['Embarked'] = df['Embarked'].fillna(df['Embarked'].mode().iloc[0])
    df['Fare'] = df['Fare'].fillna(df['Fare'].mean())

    # 数据清洗
    df.drop(['Cabin', 'Ticket', 'Name'], axis=1, inplace=True)
    df['Sex'] = df['Sex'].map({'male': 0, 'female': 1})
    df['Embarked'] = df['Embarked'].map({'S': 0, 'C': 1, 'Q': 2})

    # 特征选择
    features = ['Pclass', 'Sex', 'Age', 'SibSp', 'Parch', 'Fare', 'Embarked']
    x = df[features]
    y = df['Survived']
    return x, y

x_train, y_train = data_preprocessing(train_data)
x_test, y_test = data_preprocessing(test_data)

二、特征分析

1、数据中包含12个特征，Survived为目标变量

2、Sex和Pclass对Survived的影响较大

3、Cabin、Name和Ticket对Survived的影响较小

4、SibSp和Parch这两个特征可以合并为一个特征Family_Size

5、Fare和Age可以分别按照数值大小和区间进行分组

# 合并SibSp和Parch
train_data['Family_Size'] = train_data['SibSp'] + train_data['Parch'] + 1
test_data['Family_Size'] = test_data['SibSp'] + test_data['Parch'] + 1

# Fare按照数值大小分组
train_data['Fare_Group'] = pd.cut(x=train_data['Fare'], bins=[0, 20, 40, 60, 80, 1000])
test_data['Fare_Group'] = pd.cut(x=test_data['Fare'], bins=[0, 20, 40, 60, 80, 1000])

# Age按照区间分组
train_data['Age_Group'] = pd.cut(x=train_data['Age'], bins=[0, 12, 18, 30, 50, 100])
test_data['Age_Group'] = pd.cut(x=test_data['Age'], bins=[0, 12, 18, 30, 50, 100])

# 特征分析
sns.barplot(x='Sex', y='Survived', data=train_data)
plt.show()

sns.barplot(x='Pclass', y='Survived', data=train_data)
plt.show()

sns.barplot(x='Family_Size', y='Survived', data=train_data)
plt.show()

sns.barplot(x='Fare_Group', y='Survived', data=train_data)
plt.show()

sns.barplot(x='Age_Group', y='Survived', data=train_data)
plt.show()

三、模型训练

1、使用逻辑回归、KNN、决策树、随机森林、GBDT等常见模型进行分类预测

2、使用交叉验证法选取最优模型，并进行模型融合

from sklearn.linear_model import LogisticRegression
from sklearn.neighbors import KNeighborsClassifier
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier
from sklearn.model_selection import cross_val_score
from sklearn.ensemble import VotingClassifier

# 模型定义
models = [
    LogisticRegression(),
    KNeighborsClassifier(),
    DecisionTreeClassifier(),
    RandomForestClassifier(),
    GradientBoostingClassifier()
]

# 交叉验证找出最优模型
for model in models:
    score = cross_val_score(model, x_train, y_train, cv=5, scoring='accuracy')
    print('Model: {}, Score: {} (+/- {})'.format(model.__class__.__name__, score.mean(), score.std()))

# 模型融合
voting_classifier = VotingClassifier(estimators=[
    ('Logistic Regression', LogisticRegression()),
    ('K Nearest Neighbor', KNeighborsClassifier()),
    ('Decision Tree', DecisionTreeClassifier()),
    ('Random Forest', RandomForestClassifier()),
    ('Gradient Boosting', GradientBoostingClassifier())
], voting='soft')
voting_classifier.fit(x_train, y_train)
print('Voting Classifier:', voting_classifier.score(x_test, y_test))

四、模型调优

1、GridSearchCV进行参数调优以提高模型准确度

from sklearn.model_selection import GridSearchCV

# 随机森林分类器调优
param_grid = {
    'n_estimators': [400, 500, 600, 700, 800],
    'max_features': [3, 4, 5, 6, 7],
    'criterion': ['gini', 'entropy']
}
rfc = RandomForestClassifier(random_state=123)
grid_search = GridSearchCV(rfc, param_grid, cv=5, scoring='accuracy')
grid_search.fit(x_train, y_train)
print('Best Parameters:', grid_search.best_params_)
print('Best Score:', grid_search.best_score_)

五、模型评估

1、使用confusion_matrix、classification_report、ROC曲线等指标对模型进行评估

from sklearn.metrics import confusion_matrix, classification_report, roc_curve, auc

# 在训练数据上验证模型
y_pred_train = voting_classifier.predict(x_train)
cm_train = confusion_matrix(y_train, y_pred_train)
report_train = classification_report(y_train, y_pred_train)
print('Train Confusion Matrix:\n', cm_train)
print('Train Classification Report:\n', report_train)

# 在测试数据上验证模型
y_pred_test = voting_classifier.predict(x_test)
cm_test = confusion_matrix(y_test, y_pred_test)
report_test = classification_report(y_test, y_pred_test)
print('Test Confusion Matrix:\n', cm_test)
print('Test Classification Report:\n', report_test)

# ROC曲线和AUC值
def plot_roc_curve(model, x, y):
    y_pred = model.predict_proba(x)[:, 1]
    fpr, tpr, _ = roc_curve(y, y_pred)
    roc_auc = auc(fpr, tpr)
    plt.plot(fpr, tpr, label='ROC Curve (area = {:.2f})'.format(roc_auc))
    plt.plot([0, 1], [0, 1], 'k--')
    plt.xlim([0, 1])
    plt.ylim([0, 1.05])
    plt.xlabel('False Positive Rate')
    plt.ylabel('True Positive Rate')
    plt.title('ROC Curve')
    plt.legend(loc='lower right')
    plt.show()

plot_roc_curve(voting_classifier, x_test, y_test)