机器学习：sklearn实现心脏病预测

数据集&＃xff1a;链接&＃xff1a;https://pan.baidu.com/s/1KVRkkRp-E-W0tS4Q9qU7Ag 提取码&＃xff1a;a9wl

补充&＃xff1a;快捷显示比较图操作

离散变量

ax1 &＃61; plt.subplot(121)
ax2 &＃61; plt.subplot(122)
death_df.thal.value_counts().sort_index().plot(kind&＃61;"bar",ax &＃61; ax1)
living_df.thal.value_counts().sort_index().plot(kind&＃61;"bar",ax &＃61; ax2)


连续变量

plt.figure(figsize&＃61;(20, 10))
ax1 &＃61; plt.subplot(221)
ax2 &＃61; plt.subplot(222)ejectionFraction_groups&＃61;pd.cut(living_df["ejectionFraction"],bins&＃61;[0,0.35,0.5,0.7,0.8])
ejectionFraction_target_df &＃61; pd.concat([ejectionFraction_groups,living_df.target],axis&＃61;1)
sns.countplot(x&＃61;"ejectionFraction",hue&＃61;&＃39;target&＃39;,data&＃61;ejectionFraction_target_df,ax&＃61;ax1)ejectionFraction_groups&＃61;pd.cut(death_df["ejectionFraction"],bins&＃61;[0,0.35,0.5,0.7,0.8])
ejectionFraction_target_df &＃61; pd.concat([ejectionFraction_groups,death_df.target],axis&＃61;1)
sns.countplot(x&＃61;"ejectionFraction",hue&＃61;&＃39;target&＃39;,data&＃61;ejectionFraction_target_df,ax&＃61;ax2)


• 56.心脏病预测-数据集介绍
• 57.心脏病预测-性别与患病分析
• 58.1.心脏病预测-特征相关性分析
• 58.心脏病预测-特征预处理
• 59.心脏病预测-K近邻预测
• 60.心脏病预测-精准率召回率以及ROC曲线
• 61.心脏病预测-决策树算法评估
• 62.心脏病预测-随机森林算法评估
• 63.心脏病预测-逻辑回归算法评估
• 64.心脏病预测-SGD分类算法评估
• 65.心脏病预测-特征重要性分析

心脏病预测-数据集介绍

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as snsplt.rcParams[&＃39;font.sans-serif&＃39;] &＃61; [&＃39;SimHei&＃39;]


heart_df &＃61; pd.read_csv("./data/heart.csv")
heart_df.head()
# heart_df.info()


age - 年龄
sex - (1 &＃61; male(男性); 0 &＃61; (女性))
cp - chest pain type(胸部疼痛类型)&＃xff08;1&＃xff1a;典型的心绞痛-typical&＃xff0c;2&＃xff1a;非典型心绞痛-atypical&＃xff0c;3&＃xff1a;没有心绞痛-non-anginal&＃xff0c;4&＃xff1a;无症状-asymptomatic&＃xff09;
trestbps - 静息血压 (in mm Hg on admission to the hospital)
chol - 胆固醇 in mg/dl
fbs - (空腹血糖 > 120 mg/dl) (1 &＃61; true; 0 &＃61; false)
restecg - 静息心电图测量&＃xff08;0&＃xff1a;普通&＃xff0c;1&＃xff1a;ST-T波异常&＃xff0c;2&＃xff1a;可能左心室肥大&＃xff09;
thalach - 最高心跳率
exang - 运动诱发心绞痛 (1 &＃61; yes; 0 &＃61; no)
oldpeak - 运动相对于休息引起的ST抑制
slope - 运动ST段的峰值斜率&＃xff08;1&＃xff1a;上坡-upsloping&＃xff0c;2&＃xff1a;平的-flat&＃xff0c;3&＃xff1a;下坡-downsloping&＃xff09;
ca - 主要血管数目(0-4)
thal - 一种叫做地中海贫血的血液疾病&＃xff08;3 &＃61; normal; 6 &＃61; 固定的缺陷-fixed defect; 7 &＃61; 可逆的缺陷-reversable defect&＃xff09;
target - 是否患病 (1&＃61;yes, 0&＃61;no)


心脏病预测-性别与患病分析

# 患病的分布情况
fig,axes &＃61; plt.subplots(1,2,figsize&＃61;(10,5))
ax &＃61; heart_df.target.value_counts().plot(kind&＃61;"bar",ax&＃61;axes[0])
ax.set_title("患病分布")
ax.set_xlabel("1&＃xff1a;患病&＃xff0c;0&＃xff1a;未患病")heart_df.target.value_counts().plot(kind&＃61;"pie",autopct&＃61;"%.2f%%",labels&＃61;[&＃39;患病&＃39;,&＃39;未患病&＃39;],ax&＃61;axes[1])


# 性别和患病的分布
ax1 &＃61; plt.subplot(121)
ax &＃61; sns.countplot(x&＃61;"sex",hue&＃61;&＃39;target&＃39;,data&＃61;heart_df,ax&＃61;ax1)
ax.set_xlabel("0&＃xff1a;女性&＃xff0c;1&＃xff1a;男性")ax2 &＃61; plt.subplot(222)
heart_df[heart_df[&＃39;target&＃39;] &＃61;&＃61; 0].sex.value_counts().plot(kind&＃61;"pie",autopct&＃61;"%.2f%%",labels&＃61;[&＃39;男性&＃39;,&＃39;女性&＃39;],ax&＃61;ax2)
ax2.set_title("未患病性别比例")ax2 &＃61; plt.subplot(224)
heart_df[heart_df[&＃39;target&＃39;] &＃61;&＃61; 1].sex.value_counts().plot(kind&＃61;"pie",autopct&＃61;"%.2f%%",labels&＃61;[&＃39;男性&＃39;,&＃39;女性&＃39;],ax&＃61;ax2)
ax2.set_title("患病性别比例")


fig,axes &＃61; plt.subplots(2,1,figsize&＃61;(20,10))
sns.countplot(x&＃61;"age",hue&＃61;"target",data&＃61;heart_df,ax&＃61;axes[0])# 0-45&＃xff1a;青年人&＃xff0c;45-59&＃xff1a;中年人&＃xff0c;60-100&＃xff1a;老年人
age_type &＃61; pd.cut(heart_df.age,bins&＃61;[0,45,60,100],include_lowest&＃61;True,right&＃61;False,labels&＃61;[&＃39;青年人&＃39;,&＃39;中年人&＃39;,&＃39;老年人&＃39;])
age_target_df &＃61; pd.concat([age_type,heart_df.target],axis&＃61;1)
sns.countplot(x&＃61;"age",hue&＃61;&＃39;target&＃39;,data&＃61;age_target_df)


心脏病预测-特征相关性分析

# 统一看下所有特征的分布情况
fig,axes &＃61; plt.subplots(7,2,figsize&＃61;(10,20))
for x in range(0,14):plt.subplot(7,2,x&＃43;1)sns.distplot(heart_df.iloc[:,x],kde&＃61;True)plt.tight_layout()


plt.figure(figsize&＃61;(8,5))
sns.heatmap(heart_df.corr(),cmap&＃61;"Blues",annot&＃61;True)


心脏病预测-特征预处理

# 数据预处理
features &＃61; heart_df.drop(columns&＃61;[&＃39;target&＃39;])
targets &＃61; heart_df[&＃39;target&＃39;]


# 将离散型数据&＃xff0c;从普通的0,1,2这些&＃xff0c;转换成真正的字符串表示# sex
features.loc[features[&＃39;sex&＃39;]&＃61;&＃61;0,&＃39;sex&＃39;] &＃61; &＃39;female&＃39;
features.loc[features[&＃39;sex&＃39;]&＃61;&＃61;1,&＃39;sex&＃39;] &＃61; &＃39;male&＃39;# cp
features.loc[features[&＃39;cp&＃39;] &＃61;&＃61; 1,&＃39;cp&＃39;] &＃61; &＃39;typical&＃39;
features.loc[features[&＃39;cp&＃39;] &＃61;&＃61; 2,&＃39;cp&＃39;] &＃61; &＃39;atypical&＃39;
features.loc[features[&＃39;cp&＃39;] &＃61;&＃61; 3,&＃39;cp&＃39;] &＃61; &＃39;non-anginal&＃39;
features.loc[features[&＃39;cp&＃39;] &＃61;&＃61; 4,&＃39;cp&＃39;] &＃61; &＃39;asymptomatic&＃39;# fbs
features.loc[features[&＃39;fbs&＃39;] &＃61;&＃61; 1,&＃39;fbs&＃39;] &＃61; &＃39;true&＃39;
features.loc[features[&＃39;fbs&＃39;] &＃61;&＃61; 0,&＃39;fbs&＃39;] &＃61; &＃39;false&＃39;# exang
features.loc[features[&＃39;exang&＃39;] &＃61;&＃61; 1,&＃39;exang&＃39;] &＃61; &＃39;true&＃39;
features.loc[features[&＃39;exang&＃39;] &＃61;&＃61; 0,&＃39;exang&＃39;] &＃61; &＃39;false&＃39;# slope
features.loc[features[&＃39;slope&＃39;] &＃61;&＃61; 1,&＃39;slope&＃39;] &＃61; &＃39;true&＃39;
features.loc[features[&＃39;slope&＃39;] &＃61;&＃61; 2,&＃39;slope&＃39;] &＃61; &＃39;true&＃39;
features.loc[features[&＃39;slope&＃39;] &＃61;&＃61; 3,&＃39;slope&＃39;] &＃61; &＃39;true&＃39;# thal
features.loc[features[&＃39;thal&＃39;] &＃61;&＃61; 3,&＃39;thal&＃39;] &＃61; &＃39;normal&＃39;
features.loc[features[&＃39;thal&＃39;] &＃61;&＃61; 3,&＃39;thal&＃39;] &＃61; &＃39;fixed&＃39;
features.loc[features[&＃39;thal&＃39;] &＃61;&＃61; 3,&＃39;thal&＃39;] &＃61; &＃39;reversable&＃39;# restecg
# 0&＃xff1a;普通&＃xff0c;1&＃xff1a;ST-T波异常&＃xff0c;2&＃xff1a;可能左心室肥大
features.loc[features[&＃39;restecg&＃39;] &＃61;&＃61; 0,&＃39;restecg&＃39;] &＃61; &＃39;normal&＃39;
features.loc[features[&＃39;restecg&＃39;] &＃61;&＃61; 1,&＃39;restecg&＃39;] &＃61; &＃39;ST-T abnormal&＃39;
features.loc[features[&＃39;restecg&＃39;] &＃61;&＃61; 2,&＃39;restecg&＃39;] &＃61; &＃39;Left ventricular hypertrophy&＃39;# ca
features[&＃39;ca&＃39;].astype("object")# thal
features.thal.astype("object")features.head()


from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_splitfeatures &＃61; pd.get_dummies(features)
features_temp &＃61; StandardScaler().fit_transform(features)
# features_temp &＃61; StandardScaler().fit_transform(pd.get_dummies(features))X_train,X_test,y_train,y_test &＃61; train_test_split(features_temp,targets,test_size&＃61;0.25)


K近邻
决策树
随机森林
逻辑回归
SGD分类


心脏病预测-K近邻预测

from sklearn.neighbors import KNeighborsClassifier
from sklearn.model_selection import cross_val_score
from sklearn.metrics import precision_score,recall_score,f1_score
from sklearn.metrics import precision_recall_curve,roc_curve,average_precision_score,auc
# https://www.jianshu.com/p/c61ae11cc5f6


def plotting(estimator,y_test):fig,axes &＃61; plt.subplots(1,2,figsize&＃61;(10,5))y_predict_proba &＃61; estimator.predict_proba(X_test)precisions,recalls,thretholds &＃61; precision_recall_curve(y_test,y_predict_proba[:,1])axes[0].plot(precisions,recalls)axes[0].set_title("平均精准率&＃xff1a;%.2f"%average_precision_score(y_test,y_predict_proba[:,1]))axes[0].set_xlabel("召回率")axes[0].set_ylabel("精准率")fpr,tpr,thretholds &＃61; roc_curve(y_test,y_predict_proba[:,1])axes[1].plot(fpr,tpr)axes[1].set_title("AUC值&＃xff1a;%.2f"%auc(fpr,tpr))axes[1].set_xlabel("FPR")axes[1].set_ylabel("TPR")


# 1. K近邻
knn &＃61; KNeighborsClassifier(n_neighbors&＃61;5)
scores &＃61; cross_val_score(knn,features_temp,targets,cv&＃61;5)
print("准确率&＃xff1a;",scores.mean())knn.fit(X_train,y_train)y_predict &＃61; knn.predict(X_test)
# 精准率
print("精准率&＃xff1a;",precision_score(y_test,y_predict))
# 召回率
print("召回率&＃xff1a;",recall_score(y_test,y_predict))
# F1-Score
print("F1得分&＃xff1a;",f1_score(y_test,y_predict))plotting(knn,y_test)


心脏病预测-精准率召回率以及ROC曲线

心脏病预测-决策树算法评估

# 决策树
from sklearn.tree import DecisionTreeClassifier
tree &＃61; DecisionTreeClassifier(max_depth&＃61;10)
tree.fit(X_train,y_train)plotting(tree,y_test)


心脏病预测-随机森林算法评估

# 随机森林
from sklearn.ensemble import RandomForestClassifier
rf &＃61; RandomForestClassifier(n_estimators&＃61;100)
rf.fit(X_train,y_train)
plotting(rf,y_test)


心脏病预测-逻辑回归算法评估

# 逻辑回归
from sklearn.linear_model import LogisticRegression
logic &＃61; LogisticRegression(tol&＃61;1e-10)
logic.fit(X_train,y_train)
plotting(logic,y_test)


心脏病预测-SGD分类算法评估

# SGD分类
from sklearn.linear_model import SGDClassifier
sgd &＃61; SGDClassifier(loss&＃61;"log")
sgd.fit(X_train,y_train)
plotting(sgd,y_test)


心脏病预测-特征重要性分析

importances &＃61; pd.Series(data&＃61;rf.feature_importances_,index&＃61;features.columns).sort_values(ascending&＃61;False)
sns.barplot(y&＃61;importances.index,x&＃61;importances.values,orient&＃61;&＃39;h&＃39;)