声明：本文示例来自于GitHub用户vkasojhaa的项目，一切权利归其所有，此处仅是自己学习分享。

实现了基于机器学习的乳腺癌的恶性和良性预测，比较了不同机器学习算法之间的性能。主要目的是评估在每种算法的准确性和效率方面对数据进行分类的正确性。

基于机器学习的乳腺癌预测

代码示例

#导入依赖库
#!/usr/bin/python3
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
import seaborn as sns
import time
%matplotlib inline
#Import models from scikit learn module:
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import KFold
from sklearn.model_selection import cross_val_score
from sklearn.cross_validation import KFold
from sklearn.neighbors import KNeighborsClassifier
from sklearn.tree import DecisionTreeClassifier
from sklearn.model_selection import GridSearchCV
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import GridSearchCV
from sklearn.pipeline import Pipeline
from sklearn.svm import SVC
from sklearn import metrics

#载入数据
data = pd.read_csv("data.csv")
data.head()

#数据预处理
data.drop('id',axis=1,inplace=True) #移除id
data.drop('Unnamed: 32',axis=1,inplace=True)
print("Row, Col", data.shape)# (row,col)

#数据标记，M：恶性，B：良性
data['diagnosis'] = data['diagnosis'].map({'M':1,'B':0})
data.head()

#挖掘数据
data.describe()

探索数据

data.plot(kind='density', subplots=True, layout=(5,7), sharex=False, legend=False, fontsize=1)
plt.show()

print(data.groupby('diagnosis').size())
sns.countplot(data['diagnosis'],label="Count")
plt.show()

#划分训练集和测试集
traindf, testdf = train_test_split(data, test_size = 0.3)
labels = 'Train', 'Test'
plt.pie([70, 30], labels=labels, autopct='%1.1f%%', shadow=True)
plt.show()
print("Train set", traindf.shape)
print("Test set", testdf.shape)

features_mean= list(data.columns[1:11])
corr = data[features_mean].corr()
plt.figure(figsize=(14,14))
sns.heatmap(corr, cbar = True,  square = True, annot=True, fmt= '.2f',annot_kws={'size': 15},xticklabels= features_mean, yticklabels= features_mean,cmap= 'coolwarm')
plt.show()

模型分类

#用于模型分类和访问性能的通用函数。
Y = data['diagnosis'].values
X = data.drop('diagnosis', axis=1).values
X_train, X_test, Y_train, Y_test = train_test_split (X, Y, test_size = 0.30, random_state=21)
def classification_model(model, data, predictors, outcome):#拟合模型model.fit(data[predictors],data[outcome])#对训练集进行预测predictions = model.predict(data[predictors])#输出准确性accuracy = metrics.accuracy_score(predictions,data[outcome])print("Accuracy : %s" % "{0:.3%}".format(accuracy))#Perform k-fold cross-validation with 5 foldskfold = KFold(data.shape[0], n_folds=5)error = []for train, test in kfold:#过滤数据train_predictors = (data[predictors].iloc[train,:])# 目的在于训练算法train_target = data[outcome].iloc[train]# 使用预测变量和目标训练算法。model.fit(train_predictors, train_target)#记录每次交叉验证运行的错误error.append(model.score(data[predictors].iloc[test,:], data[outcome].iloc[test]))cv_results = cross_val_score(model, X_train, Y_train, cv=kfold, scoring='accuracy')print("Cross-Validation Score : %s" % "{0:.3%}".format(np.mean(error)))#Fit the model again so that it can be refered outside the function:model.fit(data[predictors],data[outcome])

逻辑回归模型

predictor_var = ['texture_mean','perimeter_mean','smoothness_mean','compactness_mean','symmetry_mean']
outcome_var='diagnosis'
model=LogisticRegression()
classification_model(model,traindf,predictor_var,outcome_var)

Accuracy : 91.206%
Cross-Validation Score : 90.206%

决策树模型

predictor_var = ['texture_mean','perimeter_mean','smoothness_mean','compactness_mean','symmetry_mean']
model = DecisionTreeClassifier()
classification_model(model,traindf,predictor_var,outcome_var)

Accuracy : 100.000%
Cross-Validation Score : 87.446%

predictor_var = ['texture_mean']
model = DecisionTreeClassifier()
classification_model(model,traindf,predictor_var,outcome_var)

Accuracy : 96.231%
Cross-Validation Score : 66.329%

k一近邻模型

predictor_var = ['texture_mean','perimeter_mean','smoothness_mean','compactness_mean','symmetry_mean']
model= KNeighborsClassifier()
classification_model(model,traindf,predictor_var,outcome_var)

Accuracy : 92.462%
Cross-Validation Score : 89.456%

支持向量机模型

predictor_var = ['texture_mean','perimeter_mean','smoothness_mean','compactness_mean','symmetry_mean']
model= SVC()
classification_model(model,traindf,predictor_var,outcome_var)

Accuracy : 94.472%
Cross-Validation Score : 87.937%

#几种机器学习模型的性能比较
Y = data['diagnosis'].values
X = data.drop('diagnosis', axis=1).values
X_train, X_test, Y_train, Y_test = train_test_split (X, Y, test_size = 0.30, random_state=21)
models_list = []
models_list.append(('LR', LogisticRegression()))
models_list.append(('DT', DecisionTreeClassifier()))
models_list.append(('SVM', SVC()))
models_list.append(('KNN', KNeighborsClassifier()))
num_folds = 10
results = []
names = []
for name, model in models_list:start = time.time()cv_results = cross_val_score(model, X_train, Y_train, cv=num_folds, scoring='accuracy')end = time.time()results.append(cv_results)names.append(name)print( "%s:(run time: %f)"% (name, end-start))

LR:(run time: 0.069959)
DT:(run time: 0.047665)
SVM:(run time: 0.156240)
KNN:(run time: 0.029838)

Connecting artificial intelligence (AI) with pharmaceutical sciences.

参考资料：

https://github.com/vkasojhaa/Clinical-Decision-Support-using-Machine-Learning