• 【机器学习kaggle赛事】泰坦尼克号生存预测

    目录

    写在前面

     数据集情况查看

    数据清洗

    Embarked：

    Fare

    Age

    Cabin

     特征工程

    1，探究Sex与Survived的相关性 

    2，探究Pcalss与Survived的关联性 

    3，Embarked：不同的上船地点对生存率是否有影响 

    ​4，Name与Survived的相关性

     5，Cabin与Survived之间的相关性

    6,探究孤身一人和有家人陪伴的生存率（SibSp，Parch）

     7、探究年龄Age与Survived的相关性

    8、Fare与Survived之间的相关性

    特征选择 

    构建包含最终选择特征的数据集

     分割训练数据和测试数据

    构建训练集和数据集 

    不同模型对比  

    通过决策树看各个特征的重要性

    模型训练

    写在前面

    回顾这个项目的基本流程：

    1、查看数据集，合并训练集测试集以一起进行数据清洗 

    2、数据清洗：查看数据集空缺值，并填充空缺值

    3、探索性可视化：通过透视表和图表，探究各个特征与label（需要预测的值）的相关性，若有相关性则保留。

    4、特征工程：

    • 若特征为离散型特征（类别型特征）（Name,Pclass,Embarked，Sex等），则进行独热编码，若特征为连续型数值（Fare,Age），则对特征数值进行分组，并对不同组别进行赋值，再进行Ont_hot编码处理
    • 属性几乎相同的多个特征（如SibSp: 乘客在船上的兄妹姐妹数/配偶数，Parch: 乘客在船上的父母数/子女数都属于亲人），在分别探究其与label的相关性决定保留后，可以将它们重组为新的特征。

     4、特征选择：通过相关系数来选择

     数据集情况查看

     将训练数据与测试数据连接起来，以便一起进行数据清洗。

    1. import numpy as np
    2. import pandas as pd
    3. import matplotlib.pyplot as plt
    4. import seaborn as sns
    5.  
    6. train = pd.read_csv('train.csv')
    7. test = pd.read_csv('test.csv')
    8.     
    9.     # 这里需要注意的是，如果没有后面的ignore_index=True
    10.     #那么index的值在连接后的这个新数据中是不连续的 继续从 0开始，如果要按照index删除一行数据，可能会发现多删一条。
    11. full = pd.concat([train, test], ignore_index=True)
    12. full.head()  # 默认显示5行

    PassengerId: 乘客ID；

    Survived: 生存情况，0代表不幸遇难，1代表存活；

    Pclass: 仓位等级，1为一等舱，2为二等舱，3为三等舱；

    Name: 乘客姓名；

    Sex: 性别；

    Age: 年龄；

    SibSp: 乘客在船上的兄妹姐妹数/配偶数（即同代直系亲属数）；

    Parch: 乘客在船上的父母数/子女数（即不同代直系亲属数）；

    Ticket: 船票编号；

    Fare: 船票价格；

    Cabin: 客舱号；

    Embarked: 登船港口（S: Southampton; C: Cherbourg Q: Queenstown）

    1. #查看字符型数据情况：
    2. full.describe(include=['O'])

    1. full.describe().T
    2.  
    3. #describe()函数只能查看数据类型的描述统计信息，无法查看类似字符类型的信息。
    4. #故需用info()函数进一步查看每一列的数据信息。
    5. full.info()

    数据清洗

    print(full.isnull().sum()) # 查看数据的缺失情况

    PassengerId       0
Survived        418
Pclass            0
Name              0
Sex               0
Age             263
SibSp             0
Parch             0
Ticket            0
Fare              1
Cabin          1014
Embarked          2
dtype: int64

    缺失值 Age 263， Fare 1 ，Cabin 1014， Embarked 2

    Embarked：

    分类数据，使用最常见的类别取代，用众数填充。

    1. #用众数填补Embarked    
    2. #查看众数
    3. full.Embarked.mode()
    4. full['Embarked'].fillna('S',inplace=True)

    Fare

    船票费用是类别数据，但是不同客舱的票价不同 ，用Fare空缺所在的客舱号的票价中位数来填充

    1.  
    2. #填补Fare空缺值,用Pclass==3的客舱票价的中位数来填充
    3. full[full.Fare.isnull()]#查看Fare缺失值的信息，获得Pclass
    4. full.Fare.fillna(full[full.Pclass==3]['Fare'].median(),inplace=True)

    Age

    用中位数来填充年龄
    如果是数值类型，使用平均值或者中位数进行填充

    1. #年龄(Age) 最小值为0.17，不存在0值，其数据缺失率为263/1309=20.09%，由于Age的平均数与中位数接近，故选择平均值作为缺失项的填充值。
    2. full.Age.fillna(full.Age.mean(),inplace = True)
    3. # full['Age']=full['Age'].fillna(full['Age'].mean())
    4. full.Age.describe()
    5. '''count    1309.000000
    6. mean       29.881138
    7. std        12.883193
    8. min         0.170000
    9. 25%        22.000000
    10. 50%        29.881138
    11. 75%        35.000000
    12. max        80.000000
    13. Name: Age, dtype: float64'''

    Cabin

    • 1,用U来填充，表示unkown

    字符串类型，按照实际情况填写，无法追踪的信息，用”Unknow”填充。处理Cabin缺失值 U代表Unknow

     full['Cabin'] = full['Cabin'].fillna('U')
    • 2缺失个数：

    687 超过75%的数据缺失，故不打算填补。考虑以Cabin是否缺失来构建一个新特征，看是否对生存有影响。若没有影响，则删除该列

    1. full.Cabin.isnull().sum()
    2. #1014
    3.  
    4. #创造新特征Cabin_exist，判断Cabin与Survived的相关性，相关性不大则删除
    5. full['Cabin_exist'] = full['Cabin'].map(lambda x:"Yes" if type(x)==str else "No")# 判断类型是否相等
    6. full[["Cabin_exist", "Survived"]].groupby("Cabin_exist",as_index=False).mean()
    Cabin_existSurvived
    0No0.299854
    1Yes0.666667
    1. full =full.drop('Cabin_exist',axis = 1)
    2. full.head()
    3.  
    4. #将字符转换为数值
    5. full['Cabin_exist'] = full['Cabin'].map(lambda x: 1 if type(x)==str else 0)
    6. full = full.drop('Cabin',axis = 1)
    7. full.head()
    8.  
    9. # full.loc[full.Cabin.notnull(),'Cabin'] = 1
    10. # full.loc[full.Cabin.isnull(),'Cabin'] = 0
    11. # full.Cabin.isnull().sum()#验证填充效果
    12.  
    13. sns.barplot(x="Cabin_exist",y="Survived",data=full)
    14. full.isnull().sum()
    15. #数据处理完毕
    16.  
    17. full.head()

     特征工程

    1，探究Sex与Survived的相关性 

    1. '''======================================================特征工程========================================================================'''
    2. #1，探究Sex与Survived的相关性
    3. full[['Sex','Survived']].groupby('Sex',as_index = False).mean().sort_values('Survived',ascending = False)
    4. sns.countplot(x = 'Sex',hue = 'Survived',data =full)

    1. #Sex（性别）,将字符映射到字符
    2. sex_dict = {'male': 1,'female':0}
    3. full['Sex'] = full['Sex'].map(sex_dict)
    4. full['Sex'].head()
    5.  
    6. full[["Pclass","Sex","Survived"]].groupby(["Pclass","Sex"],as_index=False).mean().sort_values(by="Survived",ascending=False)
    7.  
    8. #两维变量关系图
    9. sns.factorplot(x="Pclass",y="Survived",hue="Sex",data=full)

     

    2，探究Pcalss与Survived的关联性 

    1.  #2，探究Pcalss与Survived的关联性,相关性较高
    2. full[["Pclass","Survived"]].groupby(["Pclass"],as_index = False).mean().sort_values(by="Survived",ascending=False)
    3. sns.barplot(x="Pclass",y="Survived",data=full)

    可以看出，相关性较高，保留特征Pclass

    1. #将客舱类型进行独热编码
    2. PclassDf = pd.get_dummies(full['Pclass'],prefix = 'Pclass')
    3. PclassDf.head()
    4. #将编码后的客舱特征与原数据合并
    5. full = pd.concat([full,PclassDf],axis = 1)
    6. full.head()
    7. #删除Pclass
    8. ful = full.drop('Pclass',axis = 1)
    9. full.head()

    3，Embarked：不同的上船地点对生存率是否有影响 

    1. full[["Embarked","Survived"]].groupby("Embarked",as_index=False).count().sort_values("Survived",ascending=False)
    2. sns.barplot(x="Embarked",y="Survived",data=full)

     

    1. sns.factorplot(x="Sex",y="Survived",hue="Embarked",data=full)
    2. full[["Sex","Survived","Embarked"]].groupby(["Sex","Embarked"],as_index=False).count().sort_values("Survived",ascending=False)
    3. """S口岸，登船人数644，女性乘客占比46%；C口岸，登船人数168，女性占比接近77%；
    4. Q口岸，登船人数77，女性占比接近88%。
    5. 前面已知女性生存率明显高于男性生存率，所以上述问题可能由性别因素引起。"""

    1. #对Embarked进行独热编码
    2. EmbarkedDf = pd.get_dummies(full['Embarked'],prefix = 'Embarked')
    3. EmbarkedDf.head()
    4. # 将EmbarkedDf的特征添加至full数据集
    5. full = pd.concat([full,EmbarkedDf],axis = 1)#按列插入数据
    6. full.head()
    7. # 因为已经使用登船港口(Embarked)进行了one-hot编码产生了它的虚拟变量（dummy variables）
    8. # 所以这里把登船港口(Embarked)删掉
    9. full = full.drop('Embarked',axis = 1)
    10. full.head()

     4，Name与Survived的相关性

     从name里提取title(头衔)，地位高的人在灾难里会更有可能获得救援从而得以存活

    1. def getTitle(Name):
    2.     s1 = Name.split(',')[1]#Braund, Mr. Owen Harris	分割完取第二个元素Mr. Owen Harris
    3.     s2 = s1.split('.')[0]#得到Mr
    4.     return s2.strip( ) #strip() 方法用于移除字符串头尾指定的字符（这里是空格）
    5. full['Title'] = full['Name'].map(getTitle)#将getTitle映射给Name  得到Title
    6. full['Title'].value_counts()
    7.  
    8. full.drop('Name',axis = 1,inplace  = True)
    9. full.head()
    10.

    将头衔分为五类：

    • Officer：政府官员；
    • Royalty：王室（皇室）；
    • Mr：已婚男士；
    • Mrs：已婚妇女；
    • Miss：年轻未婚女子；
    • Master：有技能的人/教师
    1. title_dict={"Capt":"Officer","Col":"Officer","Major":"Officer","Jonkheer":"Royalty","Don":"Royalty","Sir":"Royalty","Dr":"Officer","Rev":"Officer"
    2.             ,"the Countess":"Royalty","Dona":"Royalty","Mme":"Mrs","Mlle":"Miss","Ms":"Mrs","Mr" :"Mr","Mrs" :"Mrs","Miss" :"Miss"
    3.             ,"Master" :"Master", "Lady" : "Royalty"}
    4. full['Title']=full['Title'].map(title_dict)#将title字典映射给name
    5. full.head()
    • title跟Sex有联系，联合起来分析 
    1. pd.crosstab(full.Title,full.Sex)#透视表
    2. #探索title与生存的关系
    3. full[["Title","Survived"]].groupby("Title",as_index=False).mean().sort_values("Survived",ascending=False)
    4.  
    5. sns.barplot(x="Title",y="Survived",data=full)

    full['Title'].value_counts()

    Mr         757
Miss       262
Mrs        200
Master      61
Officer     23
Royalty      6
Name: Title, dtype: int64
    1. #独热编码
    2. TitleDf = pd.get_dummies(full['Title'],prefix = 'Title')
    3. #添加进full
    4. full = pd.concat([full,TitleDf],axis = 1)
    5. full.head()
    6. #删除不需要的列
    7. full = full.drop(['Title'],axis = 1)
    8. full.head()

     5，Cabin与Survived之间的相关性

    #客场号的类别值是首字母，因此我们提取客舱号的首字母为特征
    # full['Cabin'] = full['Cabin'].map(lambda x : x[0])
    # full['Cabin'].value_counts
    # full['Cabin']=full['Cabin'].map(lambda x:x[0])
    # full['Cabin'].value_counts()
    """因为之前用处理过，所以不再提取首字母，如果用u来填充的话，可以用这种方法"""

    6,探究孤身一人和有家人陪伴的生存率（SibSp，Parch）

    1. full[['SibSp','Survived']].groupby(['SibSp'],as_index = False).mean().sort_values('Survived',ascending = False)
    2. sns.barplot(x = 'SibSp',y = 'Survived',data = full)

    1. full[["Parch","Survived"]].groupby("Parch",as_index=False).mean().sort_values("Survived",ascending=False)
    2. sns.barplot(x = 'Parch',y = 'Survived',data = full)

     可以看出SibSp，Parch 跟Survived 有相关联系，所以用SibSp，Parch构建家庭人数和家庭类别的新特征Alone，family_small，family_big

    1. # 构建家庭人数和家庭类别的新特征
    2. full['family'] = full['Parch'] + full['SibSp'] + 1#Parch，SibSp为0时，只有自己一个人，+1
    3. full['Alone'] = np.where(full['family'] == 1,1,0)
    4.  
    5. full['family_small'] = np.where((full['family']>=2) & (full['family']<=4),1,0)
    6. full['family_big'] = np.where(full['family']>=5,1,0)
    7. full.head()

     探究家庭与存活率的关系  证实孤身一人存活率不高

    1. #探究家庭与存活率的关系  证实孤身一人存活率不高
    2. full[['family','Survived']].groupby('family',as_index = False).mean().sort_values('Survived',ascending = False)
    3. sns.barplot(x = 'family',y = 'Survived',data = full)
    4.  
    5. #孤身一人对生存率是否有影响
    6. full[["Alone","Survived"]].groupby("Alone",as_index = False).mean().sort_values("Survived",ascending=False)
    7. sns.barplot(x="Alone",y="Survived",data=full)
    8.  
    9. full = full.drop('family',axis = 1)
    sns.factorplot(x="Pclass",y="Survived",hue="Alone",data=full)

     7、探究年龄Age与Survived的相关性

    1. #查看Age的分布情况
    2. sns.violinplot(y="Age",data=full)
    3. #查看生存与死亡乘客的年龄分布
    4. sns.violinplot(y="Age",x="Survived",data=full)
    5. #将年龄分为五组
    6. full['AgeCut'] = pd.cut(full.Age,5)#要用full['AgeCut']，而不是full.AgeCut,这样AgeCut才会在Index里
    7. full.AgeCut.value_counts().sort_index()
    8.  
    9. full[['AgeCut','Survived']].groupby('AgeCut',as_index = False).mean().sort_values('Survived',ascending = False)
    (0.0902, 16.136]    134
(16.136, 32.102]    787
(32.102, 48.068]    269
(48.068, 64.034]    106
(64.034, 80.0]       13
Name: AgeCut, dtype: int64

     根据各个分段，重新给Age赋值并进行独热编码

    1. #根据各个分段，重新给Age赋值
    2. full.loc[full.Age <= 16.136,'Age'] = 1
    3. full.loc[(full.Age > 16.136)&(full.Age <=32.102),'Age'] = 2
    4. full.loc[(full.Age > 32.102)&(full.Age <=48.068),'Age'] = 3
    5. full.loc[(full.Age > 48.068)&(full.Age <=64.034),'Age'] = 4
    6. full.loc[full.Age > 64.034,'Age'] = 5
    7. full.head()
    8.  
    9. AgeDf = pd.get_dummies(full['Age'],prefix =' Age')
    10. full = pd.concat([full,AgeDf],axis = 1)
    11. full = full.drop(['Age','AgeCut'],axis = 1)
    12. full.head()

    8、Fare与Survived之间的相关性

    1. sns.violinplot(y="Fare",data=train)
    2. #对比生死乘客的票价
    3. sns.violinplot(y="Fare",x="Survived",data=train) 

    1. # 当然这里也可以用seaborn的displot进行绘制，但是displot的纵坐标是比率，hist的纵坐标是实际个数count；
    2. # figsize调整画布大小
    3. full['Fare'].hist(color='green', bins=30, figsize=(8,4))

    1. #分组
    2. full['FareCut'] = pd.cut(full.Fare,5)
    3. full.FareCut.value_counts().sort_index()
    4. # full.head()
    5.  
    6. full[['FareCut','Survived']].groupby('FareCut',as_index = False).mean().sort_values('Survived',ascending = False)
    7. #重新赋值
    8. full.loc[full.Fare<=7.854,'Fare']=1
    9. full.loc[(full.Fare>7.854)&(full.Fare<=10.5),'Fare']=2
    10. full.loc[(full.Fare>10.5)&(full.Fare<=21.558),'Fare']=3
    11. full.loc[(full.Fare>21.558)&(full.Fare<=41.579),'Fare']=4
    12. full.loc[full.Fare>41.579,'Fare']=5
    13. full.head()
    14.  
    15. #进行独热编码
    16. FareDf = pd.get_dummies(full['Fare'],prefix = 'Fare')
    17. full = pd.concat([full,FareDf],axis = 1)
    18. full = full.drop(['Fare','FareCut'],axis = 1)
    19. full.head()
    20.  
    21. full = full.drop(['SibSp','Parch','Pclass'],axis = 1)
    22. full.head()

    特征选择 

    1. """=============================================特征选择和特征降维========================================================="""
    2. full.info()
    3.  
    4.  
    5. corr_df=full.corr()  
    6. corr_df
    7.  
    8. #用图形直观查看线性相关系数
    9. plt.figure(figsize=(16,16))
    10. plt.title("Pearson Correlation of Features")
    11. sns.heatmap(corr_df,linewidths=0.1,square=True,linecolor="white",annot=True,cmap='YlGnBu',vmin=-1,vmax=1)
    12.  
    13. corr_df['Survived'].sort_values(ascending = False)
    Survived         1.000000
Title_Mrs        0.344935
Title_Miss       0.332795
Cabin_exist      0.316912
Pclass_1         0.285904
family_small     0.279855
Fare_5.0         0.266217
Embarked_C       0.168240
 Age_1.0         0.121485
Pclass_2         0.093349
Title_Master     0.085221
Fare_4.0         0.058052
Fare_3.0         0.043153
Title_Royalty    0.033391
 Age_4.0         0.030350
 Age_3.0         0.021711
Embarked_Q       0.003650
Title_Officer   -0.031316
 Age_5.0        -0.067344
 Age_2.0        -0.097245
family_big      -0.125147
Embarked_S      -0.149683
Fare_1.0        -0.164287
Fare_2.0        -0.198067
Alone           -0.203367
Pclass_3        -0.322308
Sex             -0.543351
Title_Mr        -0.549199
Name: Survived, dtype: float64

    构建包含最终选择特征的数据集

     最终选择特征如下：

    1. full_x=pd.concat([TitleDf,PclassDf,EmbarkedDf,FareDf,AgeDf,full['Sex'],full['family_small'],full['Cabin_exist']
    2.                   ,full['family_big'],full['Alone']],axis=1)
    3. full_x.head()

     分割训练数据和测试数据

    1. ​
    2. #前891行为原始训练数据
    3. source_x=full_x.loc[0:890,:]#提取特征值
    4. source_y=full.loc[0:890,'Survived']#提取标签值
    5.  
    6. #后418行是test,预测数据
    7. pred_x=full_x.loc[891:,:]
    8. source_x.shape#(891, 23)
    9. source_y.shape#(891,)
    10. pred_x.shape#(418, 23)
    11.  
    12. ​

    构建训练集和数据集 

    1. #训练数据集和测试数据集，按照二八原则分为训练数据和测试数据，其中80%为训练数据
    2. from sklearn.model_selection import train_test_split
    3. x_train,x_test,y_train,y_test =train_test_split(source_x,source_y,train_size=0.8)
    4. print('训练数据集特征:{0},训练数据集标签:{1}'.format(x_train.shape,y_train.shape))
    5. print('测试数据集特征:{0},测试数据集标签:{1}'.format(x_test.shape,y_test.shape))
    6.  
    7.  
    8. #训练数据集特征:(712, 27),训练数据集标签:(712,)
    9. #测试数据集特征:(179, 27),测试数据集标签:(179,)
    1. #对x_train,x_test进行标准化
    2. from sklearn.preprocessing import StandardScaler
    3. sc = StandardScaler()
    4. x_train_std=sc.fit_transform(x_train)
    5. x_test_std=sc.transform(x_test)

    不同模型对比  

    1. from sklearn.model_selection import cross_val_score
    2.  
    3. from sklearn.neighbors import KNeighborsClassifier
    4. from sklearn.naive_bayes import GaussianNB
    5. from sklearn.tree import DecisionTreeClassifier
    6. from sklearn.ensemble import RandomForestClassifier
    7. from sklearn.ensemble import GradientBoostingClassifier
    8. from sklearn.svm import SVC
    9.  
    10. models=[KNeighborsClassifier(),GaussianNB(),DecisionTreeClassifier(),RandomForestClassifier(),
    11.        GradientBoostingClassifier(),SVC()]
    12.  
    13. # 计算各模型得分
    14. names=['KNN','NB','Tree','RF','GDBT','SVM']
    15. for name, model in zip(names,models):
    16.     score=cross_val_score(model,Xa,y,cv=5)
    17.     print("{}:{},{}".format(name,score.mean(),score))
    KNN:0.8159374803841566,[0.81564246 0.76404494 0.84269663 0.81460674 0.84269663]
NB:0.7385412089636557,[0.69832402 0.74719101 0.82022472 0.57303371 0.85393258]
Tree:0.8069612704789405,[0.80446927 0.76404494 0.85393258 0.79775281 0.81460674]
RF:0.8125666938673028,[0.81564246 0.75842697 0.8258427  0.82022472 0.84269663]
GDBT:0.8092147385600402,[0.79888268 0.7752809  0.83146067 0.78651685 0.85393258]
SVM:0.821542903772519,[0.82681564 0.80337079 0.8258427  0.78651685 0.86516854]
    1. # 使用标准化的数据 scaled data
    2. names=['KNN','NB','Tree','RF','GDBT','SVM']
    3. for name, model in zip(names,models):
    4.     score=cross_val_score(model,X_std,y,cv=5)
    5.     print("{}:{},{}".format(name,score.mean(),score))
    KNN:0.8204569706860838,[0.79329609 0.76404494 0.87640449 0.82022472 0.84831461]
NB:0.7216872763793861,[0.69832402 0.74719101 0.82022472 0.48876404 0.85393258]
Tree:0.8092084614901763,[0.80446927 0.76404494 0.84831461 0.79775281 0.83146067]
RF:0.8226727763480006,[0.82122905 0.76966292 0.85955056 0.81460674 0.84831461]
GDBT:0.8103383340656581,[0.79888268 0.7752809  0.83146067 0.78651685 0.85955056]
SVM:0.8170610758897746,[0.81564246 0.80337079 0.8258427  0.78089888 0.85955056]

    通过决策树看各个特征的重要性

    1. clf = DecisionTreeClassifier(criterion = 'entropy',random_state = 30,splitter = 'random')
    2. clf.fit(x_train_std,y_train)
    3. score = clf.score(x_test_std,y_test)
    4. score
    5.  
    6.  
    7. fi=pd.DataFrame({'importance':clf.feature_importances_},index=X.columns)
    8. fi.sort_values('importance',ascending=False)

     

    1. fi.sort_values('importance',ascending=False).plot.bar(figsize=(11,7))
    2. plt.xticks(rotation=30)
    3. plt.title('Feature Importance',size='x-large')
    1.  
    2. from sklearn.model_selection import GridSearchCV
    3.  
    4. param_grid={'n_neighbors':[1,2,3,4,5,6,7,8,9]}
    5. grid_search=GridSearchCV(KNeighborsClassifier(),param_grid,cv=5)
    6.  
    7. grid_search.fit(x_train_std,y_train)
    8.  
    9. grid_search.best_params_,grid_search.best_score_
    1. #LogisticRegression
    2. param_grid={'C':[0.01,0.1,1,10]}
    3. grid_search=GridSearchCV(LogisticRegression(),param_grid,cv=5)
    4.  
    5. grid_search.fit(x_train_std,y_train)
    6.  
    7. grid_search.best_params_,grid_search.best_score_
    8.  
    9. # second round grid search
    10. param_grid={'C':[0.04,0.06,0.08,0.1,0.12,0.14]}
    11. grid_search=GridSearchCV(LogisticRegression(),param_grid,cv=5)
    12.  
    13. grid_search.fit(x_train_std,y_train)
    14.  
    15. grid_search.best_params_,grid_search.best_score_
    1. #Support Vector Machine
    2.  
    3. param_grid={'C':[0.01,0.1,1,10],'gamma':[0.01,0.1,1,10]}
    4. grid_search=GridSearchCV(SVC(),param_grid,cv=5)
    5.  
    6. grid_search.fit(x_train_std,y_train)
    7.  
    8. grid_search.best_params_,grid_search.best_score_
    9.  
    10.  
    11. #second round grid search
    12. param_grid={'C':[2,4,6,8,10,12,14],'gamma':[0.008,0.01,0.012,0.015,0.02]}
    13. grid_search=GridSearchCV(SVC(),param_grid,cv=5)
    14.  
    15. grid_search.fit(x_train_std,y_train)
    16.  
    17. grid_search.best_params_,grid_search.best_score_
    1. #Gradient Boosting Decision Tree
    2.  
    3. param_grid={'n_estimators':[30,50,80,120,200],'learning_rate':[0.05,0.1,0.5,1],'max_depth':[1,2,3,4,5]}
    4. grid_search=GridSearchCV(GradientBoostingClassifier(),param_grid,cv=5)
    5.  
    6. grid_search.fit(x_train_std,y_train)
    7.  
    8. grid_search.best_params_,grid_search.best_score_
    9.  
    10. #second round grid search
    11. param_grid={'C':[2,4,6,8,10,12,14],'gamma':[0.008,0.01,0.012,0.015,0.02]}
    12. grid_search=GridSearchCV(SVC(),param_grid,cv=5)
    13.  
    14. grid_search.fit(x_train_std,y_train)
    15.  
    16. grid_search.best_params_,grid_search.best_score_
    1. #Gradient Boosting Decision Tree
    2. param_grid={'n_estimators':[30,50,80,120,200],'learning_rate':[0.05,0.1,0.5,1],'max_depth':[1,2,3,4,5]}
    3. grid_search=GridSearchCV(GradientBoostingClassifier(),param_grid,cv=5)
    4.  
    5. grid_search.fit(x_train_std,y_train)
    6.  
    7. grid_search.best_params_,grid_search.best_score_
    8.  
    9.  
    10. #second round search
    11. param_grid={'n_estimators':[100,120,140,160],'learning_rate':[0.05,0.08,0.1,0.12],'max_depth':[3,4]}
    12. grid_search=GridSearchCV(GradientBoostingClassifier(),param_grid,cv=5)
    13.  
    14. grid_search.fit(x_train_std,y_train)
    15.  
    16. grid_search.best_params_,grid_search.best_score_

    模型训练

    1. #逻辑回归
    2. from sklearn.linear_model import LogisticRegression
    3. model1 = LogisticRegression()
    4. model1.fit(x_train_std,y_train)
    5.  
    6. LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,
    7.               intercept_scaling=1, max_iter=100, multi_class='ovr', n_jobs=1,
    8.               penalty='l2', random_state=None, solver='liblinear', tol=0.0001,
    9.               verbose=0, warm_start=False)
    10.  
    11. #在测试集上得出模型正确率
    12. model1.score(x_test_std,y_test)
    13.  
    14. 0.8212290502793296
    15.  
    16.  
    17. pred1 = model1.predict(x_test)
    18. score1 = model1.score(x_train,y_train)
    19. score1
    20.  
    21.  
    22. #使用训练得到的模型对pred_x的生存情况进行预测
    23. pred_x_std=sc.fit_transform(pred_x)
    24. pred_y=model1.predict(pred_x_std)
    25. pred_y
    26.

    泰坦尼克号生存预测 (Logistic and KNN)

    kaggle 泰坦尼克号生存预测——六种算法模型实现与比较

    GDBT特征重要性可视化

    fit_transform,fit,transform区别和作用详解

    sklearn中的cross_val_score()函数参数

    train_test_split用法

    pandas 读取或者选择某几列                       模型调参利器 gridSearchCV（网格搜索）

  • 相关阅读:
    在k8s中部署高可用程序实践和资源治理
    参考线平滑-FemPosDeviation-SQP
    黑猫带你学Makefile第1篇：什么是Makefile
    【EMC专题】电磁兼容学科的发展
    文心一言初体验，和ChatGPT语言理解能力比较
    蓝桥杯（等差素数列，C++）
    容器化｜自建 MySQL 集群迁移到 Kubernetes
    【嵌入式项目应用】__一款简单易用的嵌入式软件程序框架_时间片轮询框架
    SpringCloudGateway集成SpringDoc
    Linux 终端与进程
  • 原文地址：https://blog.csdn.net/m0_51933492/article/details/126895547