预测脑中风

中风是危害人们健康的重大疾病之一，根据医学杂志《柳叶刀》发表的一篇文章中的数据，中风在我国居民死亡原因中高居第一位，而且我国居民的中风的比例也是世界上最高的之一。因此，了解居民的身体状况与出现中风的联系，进而分析中风的成因，实现对不同的人群，有针对性的采取预防的措施，是有积极的意义的。

附件1中随机搜集了5000多个人的数据，请你根据附件1中的信息，作以下分析：

数据中列举的各因素中，哪些对中风的影响较大？给出你的排序，并分析各因素的不同取值与中风发病率的关系。
利用这些数据，寻找出几个最容易出现中风的小的群体，(一般人数不超过总人数的5%), 分析这些群体的个体特征。
建立数学模型，对不同的人分析其中风的可能性大小。并利用你的模型对附件2中的10个人按中风可能性大小排序。

from google.colab import drive
drive.mount('/content/drive')

Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount("/content/drive", force_remount=True).

%cd /content/drive/My Drive
!pwd

/content/drive/My Drive
/content/drive/My Drive

import matplotlib.font_manager as mfm
font_path="/content/drive/My Drive/Colab Notebooks/simhei.ttf"
prop=mfm.FontProperties(fname=font_path)

%cd ./Colab Notebooks/XXXXX
!pwd

/content/drive/My Drive/Colab Notebooks/XXXXX
/content/drive/My Drive/Colab Notebooks/XXXXX

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as snsimport warnings
warnings.filterwarnings("ignore")

data=pd.read_csv('附件1.csv')#读取
excel2=pd.read_excel('附件2.xlsx')

backup=data.copy()#备份
excel1=data
print('查看病例数据的大小:',excel1.shape)
length_excel1=excel1.shape[0]

查看病例数据的大小: (5110, 12)

print('查看病例数据的变量信息:')#观察有多少缺少值
print(excel1.info())

查看病例数据的变量信息:
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 5110 entries, 0 to 5109
Data columns (total 12 columns):#   Column             Non-Null Count  Dtype  
---  ------             --------------  -----  0   id                 5110 non-null   int64  1   gender             5110 non-null   object 2   age                5110 non-null   float643   hypertension       5110 non-null   int64  4   heart_disease      5110 non-null   int64  5   ever_married       5110 non-null   object 6   work_type          5110 non-null   object 7   Residence_type     5110 non-null   object 8   avg_glucose_level  5110 non-null   float649   bmi                4909 non-null   float6410  smoking_status     5110 non-null   object 11  stroke             5110 non-null   int64  
dtypes: float64(3), int64(4), object(5)
memory usage: 479.2+ KB
None

print("每个病例数据的内容分布")#观察每项数据有多少个
print(excel1['gender'].value_counts(),'\n\n')
print(excel1['age'].value_counts(),'\n\n')
print(excel1['hypertension'].value_counts(),'\n\n')
print(excel1['heart_disease'].value_counts(),'\n\n')
print(excel1['ever_married'].value_counts(),'\n\n')
print(excel1['work_type'].value_counts(),'\n\n')
print(excel1['Residence_type'].value_counts(),'\n\n')
print(excel1['avg_glucose_level'].value_counts(),'\n\n')
print(excel1['bmi'].value_counts(),'\n\n')
print(excel1['smoking_status'].value_counts(),'\n\n')
print(excel1['stroke'].value_counts())

每个病例数据的内容分布
Female    2994
Male      2115
Other        1
Name: gender, dtype: int64 78.00    102
57.00     95
52.00     90
54.00     87
51.00     86... 
1.40       3
0.48       3
0.16       3
0.40       2
0.08       2
Name: age, Length: 104, dtype: int64 0    4612
1     498
Name: hypertension, dtype: int64 0    4834
1     276
Name: heart_disease, dtype: int64 Yes    3353
No     1757
Name: ever_married, dtype: int64 Private          2925
Self-employed     819
children          687
Govt_job          657
Never_worked       22
Name: work_type, dtype: int64 Urban    2596
Rural    2514
Name: Residence_type, dtype: int64 93.88     6
91.68     5
91.85     5
83.16     5
73.00     5..
111.93    1
94.40     1
95.57     1
66.29     1
85.28     1
Name: avg_glucose_level, Length: 3979, dtype: int64 28.7    41
28.4    38
26.7    37
27.6    37
26.1    37..
48.7     1
49.2     1
51.0     1
49.4     1
14.9     1
Name: bmi, Length: 418, dtype: int64 never smoked       1892
Unknown            1544
formerly smoked     885
smokes              789
Name: smoking_status, dtype: int64 0    4861
1     249
Name: stroke, dtype: int64

print('查看病例数据的数值变量的描述性分析:')#对于数值型变量，观察数值范围
print(excel1['age'].describe(),'\n\n')
print(excel1['avg_glucose_level'].describe(),'\n\n')
print(excel1['bmi'].describe(),'\n\n')

查看病例数据的数值变量的描述性分析:
count    5110.000000
mean       43.226614
std        22.612647
min         0.080000
25%        25.000000
50%        45.000000
75%        61.000000
max        82.000000
Name: age, dtype: float64 count    5110.000000
mean      106.147677
std        45.283560
min        55.120000
25%        77.245000
50%        91.885000
75%       114.090000
max       271.740000
Name: avg_glucose_level, dtype: float64 count    4909.000000
mean       28.893237
std         7.854067
min        10.300000
25%        23.500000
50%        28.100000
75%        33.100000
max        97.600000
Name: bmi, dtype: float64

excel1 = excel1.fillna(np.mean(excel1['bmi']))#bmi数据有遗漏，用平均值填充变量
excel1.drop(excel1[excel1['gender'] == 'Other'].index, inplace= True)print('查看附件1病例数据大小:',excel1.shape)
print('查看附件2预测数据大小:',excel2.shape)
length_excel1=excel1.shape[0]
length_excel2=excel2.shape[0]

查看附件1病例数据大小: (5109, 12)
查看附件2预测数据大小: (10, 12)

#把两个数据放一起处理
all_data = pd.concat((excel1, excel2)).reset_index(drop=True)#合并
all_data.drop(columns=['id'],axis=1,inplace=True)#丢弃id
all_data.drop(columns=['gender'],axis=1,inplace=True)#丢弃id
print('查看去掉id列后所有数据的大小:',all_data.shape)

查看去掉id列后所有数据的大小: (5119, 10)

all_data[['hypertension', 'heart_disease', 'stroke']] = all_data[['hypertension', 'heart_disease', 'stroke']].astype(str)#改变数据类型 int转object
print('查看所有变量信息:')
print(all_data.info())

查看所有变量信息:
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 5119 entries, 0 to 5118
Data columns (total 10 columns):#   Column             Non-Null Count  Dtype  
---  ------             --------------  -----  0   age                5119 non-null   float641   hypertension       5119 non-null   object 2   heart_disease      5119 non-null   object 3   ever_married       5119 non-null   object 4   work_type          5119 non-null   object 5   Residence_type     5119 non-null   object 6   avg_glucose_level  5119 non-null   float647   bmi                5119 non-null   float648   smoking_status     5119 non-null   object 9   stroke             5119 non-null   object 
dtypes: float64(3), object(7)
memory usage: 400.0+ KB
None

all_data=pd.get_dummies(all_data, drop_first=True)#字符串变量整形处理#把不同的变量内容转化为0，1分布#哑变量处理from sklearn.preprocessing import StandardScaler
#s = StandardScaler()
all_data[['bmi', 'avg_glucose_level', 'age']] = StandardScaler().fit_transform(all_data[['bmi', 'avg_glucose_level', 'age']])#特征标准化#object转uint8print('\n查看特征标准化后所有变量信息:')
print(all_data.info())

查看特征标准化后所有变量信息:
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 5119 entries, 0 to 5118
Data columns (total 16 columns):#   Column                          Non-Null Count  Dtype  
---  ------                          --------------  -----  0   age                             5119 non-null   float641   avg_glucose_level               5119 non-null   float642   bmi                             5119 non-null   float643   hypertension_1                  5119 non-null   uint8  4   heart_disease_1                 5119 non-null   uint8  5   ever_married_Yes                5119 non-null   uint8  6   work_type_Never_worked          5119 non-null   uint8  7   work_type_Private               5119 non-null   uint8  8   work_type_Self-employed         5119 non-null   uint8  9   work_type_children              5119 non-null   uint8  10  Residence_type_Urban            5119 non-null   uint8  11  smoking_status_formerly smoked  5119 non-null   uint8  12  smoking_status_never smoked     5119 non-null   uint8  13  smoking_status_smokes           5119 non-null   uint8  14  stroke_1.0                      5119 non-null   uint8  15  stroke_nan                      5119 non-null   uint8  
dtypes: float64(3), uint8(13)
memory usage: 185.1 KB
None

excel1=all_data[:length_excel1]
excel2=all_data[length_excel1:]
print('最终病例集数量:',excel1.shape)
print('最终预测集数量:',excel2.shape)

最终病例集数量: (5109, 16)
最终预测集数量: (10, 16)

X=excel1.drop(['stroke_1.0'],axis=1)
Y=excel1['stroke_1.0']
z=excel2.drop(['stroke_1.0'],axis=1)

from sklearn.model_selection import train_test_splitx_train,x_test,y_train,y_test=train_test_split(X,Y,test_size=0.2,random_state=1024)
print('查看划分的训练集的组成:\n',y_train.value_counts())
print('查看划分的测试集的组成:\n',y_test.value_counts())

查看划分的训练集的组成:0    3881
1     206
Name: stroke_1.0, dtype: int64
查看划分的测试集的组成:0    979
1     43
Name: stroke_1.0, dtype: int64


from imblearn.over_sampling import RandomOverSampler
over=RandomOverSampler(sampling_strategy=1)
x_smote,y_smote=over.fit_resample(x_train,y_train)
print('查看过采样的训练集的组成:\n',y_smote.value_counts())
print('查看过采样的测试集的组成:\n',y_test.value_counts())

查看过采样的训练集的组成:0    3881
1    3881
Name: stroke_1.0, dtype: int64
查看过采样的测试集的组成:0    979
1     43
Name: stroke_1.0, dtype: int64

'''
from imblearn.under_sampling import RandomUnderSampler
over=RandomUnderSampler(random_state=42)
x_smote,y_smote=over.fit_resample(x_train,y_train)
print('查看过采样的训练集的组成:\n',y_smote.value_counts())
print('查看过采样的测试集的组成:\n',y_test.value_counts())
'''

"\nfrom imblearn.under_sampling import RandomUnderSampler\nover=RandomUnderSampler(random_state=42)\nx_smote,y_smote=over.fit_resample(x_train,y_train)\nprint('查看过采样的训练集的组成:\n',y_smote.value_counts())\nprint('查看过采样的测试集的组成:\n',y_test.value_counts())\n"

x_train=x_smote
y_train=y_smote

from sklearn.metrics import accuracy_score,roc_auc_score,classification_report,confusion_matrix,roc_curvedef model_choose(names, models, colors, pres):plt.figure(figsize=(5,5))#设置画布大小#plt.rcParams['font.sans-serif'] =['SimHei']#设置字体for (name, model, color) in zip(names, models, colors):model.fit(x_train, y_train)#训练模型y_pre=model.predict(x_test)#得预测值print('\n\n')print(name,'Accuracy:',accuracy_score(y_test,y_pre))#计算准确率score=model.predict_proba(x_test)[:,1]print(name,'AUC:', roc_auc_score(y_test,score))#计算AUCprint(classification_report(y_test, y_pre))#其它数值计算fpr,tpr,thresholds=roc_curve(y_test, score)#画ROC曲线plt.plot(fpr, tpr, color=color, lw=1, label=f'{name} AUC= {roc_auc_score(y_test,score)}')plt.axis('square')plt.xlabel('False Positive Rate') plt.ylabel('True Positive Rate') plt.title('各类模型ROC曲线对比',fontproperties=prop) plt.legend(loc="lower right") plt.savefig('各类型模型ROC曲线对比')plt.show()countnum=0for (name, model, color, z_pre) in zip(names, models, colors, pres):model.fit(x_train, y_train)y_pre=model.predict(x_test)#得预测值z_pre=model.predict_proba(z)[:,1]#对附件二的十个人进行预测概率model_pre[countnum]=z_precountnum+=1#print(z_pre)cm=confusion_matrix(y_test,y_pre)#画混淆矩阵/热力图ax=sns.heatmap(cm,cmap='Pastel1',annot=True,fmt='g',xticklabels=['0','1'],yticklabels=['0','1'])ax.set_title(name)ax.set_xlabel('predict')ax.set_ylabel('true')plt.savefig(name)plt.show()#逻辑回归
from sklearn.linear_model import LogisticRegression
model1=LogisticRegression()
#k近邻
from sklearn.neighbors import KNeighborsClassifier
model2=KNeighborsClassifier(n_neighbors=10)
#决策树
from sklearn.tree import DecisionTreeClassifier
model3=DecisionTreeClassifier()
#随机森林
from sklearn.ensemble import RandomForestClassifier
model4=RandomForestClassifier(n_estimators=300)
#向量机
from sklearn.svm import SVC
model5=SVC(kernel='rbf', probability=True)
#梯度下降
from sklearn.linear_model import SGDClassifier
model6=SGDClassifier(random_state=12, loss='log')
#梯度上升
from sklearn.ensemble import GradientBoostingClassifier
model7=GradientBoostingClassifier(random_state=6)

model_name=['Logistic','KNeighbors','DecisionTree','RandomForest','SVC','SGD','GradientBoosting']
model_type=[model1,model2,model3,model4,model5,model6,model7]
model_color = ['yellow', 'aqua', 'lawngreen', 'fuchsia', 'orange', 'r', 'pink']
model_pre=[np.array([]),np.array([]),np.array([]),np.array([]),np.array([]),np.array([]),np.array([]),]model_choose(model_name, model_type, model_color, model_pre)

Logistic Accuracy: 0.7514677103718199
Logistic AUC: 0.8284438321020501precision    recall  f1-score   support0       0.99      0.75      0.85       9791       0.12      0.79      0.21        43accuracy                           0.75      1022macro avg       0.55      0.77      0.53      1022
weighted avg       0.95      0.75      0.83      1022KNeighbors Accuracy: 0.8209393346379648
KNeighbors AUC: 0.7010832125804689precision    recall  f1-score   support0       0.97      0.84      0.90       9791       0.11      0.47      0.18        43accuracy                           0.82      1022macro avg       0.54      0.65      0.54      1022
weighted avg       0.94      0.82      0.87      1022DecisionTree Accuracy: 0.9334637964774951
DecisionTree AUC: 0.5428177779889303precision    recall  f1-score   support0       0.96      0.97      0.97       9791       0.14      0.12      0.13        43accuracy                           0.93      1022macro avg       0.55      0.54      0.55      1022
weighted avg       0.93      0.93      0.93      1022RandomForest Accuracy: 0.9461839530332681
RandomForest AUC: 0.7939520630923819precision    recall  f1-score   support0       0.96      0.98      0.97       9791       0.17      0.07      0.10        43accuracy                           0.95      1022macro avg       0.56      0.53      0.54      1022
weighted avg       0.93      0.95      0.94      1022SVC Accuracy: 0.7436399217221135
SVC AUC: 0.7864218352851747precision    recall  f1-score   support0       0.98      0.75      0.85       9791       0.11      0.70      0.19        43accuracy                           0.74      1022macro avg       0.55      0.72      0.52      1022
weighted avg       0.95      0.74      0.82      1022SGD Accuracy: 0.7534246575342466
SGD AUC: 0.8246668408675202precision    recall  f1-score   support0       0.99      0.75      0.85       9791       0.12      0.77      0.21        43accuracy                           0.75      1022macro avg       0.55      0.76      0.53      1022
weighted avg       0.95      0.75      0.83      1022GradientBoosting Accuracy: 0.8062622309197651
GradientBoosting AUC: 0.8325533886025133precision    recall  f1-score   support0       0.98      0.81      0.89       9791       0.14      0.70      0.23        43accuracy                           0.81      1022macro avg       0.56      0.75      0.56      1022
weighted avg       0.95      0.81      0.86      1022

result=(model_pre[0]+model_pre[3]+model_pre[5]+model_pre[6])/4#取效果好的四个模型计算
print(result)#10个人的概率
print(np.sort(result))#概率排序
print(np.argsort(-result)+1)#输出原索引

[0.77866337 0.29122944 0.86967168 0.62082311 0.73240077 0.130646320.78216716 0.28405546 0.3040807  0.57776553]
[0.13064632 0.28405546 0.29122944 0.3040807  0.57776553 0.620823110.73240077 0.77866337 0.78216716 0.86967168]
[ 3  7  1  5  4 10  9  2  8  6]