机器学习过程中常用的API总结

（1）决策树

from sklearn.feature_extraction import DictVectorizer
from sklearn.preprocessing import LabelBinarizer
from sklearn import tree

vec = DictVectorizer()#实例化特征变量
feature_name = vec.get_feature_names()
lb = LabelBinarizer()#实例化标记变量
dummyY = lb.fit_transform(list)
dummyX = vec.fit_transform(list)
clf = tree.DecisionTreeClassifier(critetion='entropy')
clf.fit(dummyX,dummyY)

#可视化过程
f = tree.export_graphviz(clf,feature_names,outfile)
#命令行输入dot -Tpdf file.dot -o output.pdf

(2)KNN

import sklearn import neighbors
knn = neighbors.KneighborsClassifier()

from sklearn.metrics import confusion_matrix
from sklearn.metrics import classifier_report
confusion_matrix() #混淆矩阵
classifier_report()#构建显示分类指标，如精确率，召回率，F1的文本报告

精确率（precision）=预测正确的数目/预测值中出现该类型的数目
召回率(recall)=预测正确的数目/总数目

F1=pre*recall/(pre+recall)

(3)SLR

sklearn实现法

from sklearn import linear_model
regr = linear_model.LinearRegression()

#单一变量线性回归模型的建立，传入参数形式必须是list[list]
X = [[1],[2],[3]]
Y = [7，8，9]
regr.fit(X,Y)

numpy实现

import numpy as np

x = [1,2,3]

y = [5,6,7]

coef = np.polyfit(x,y,1)#返回多项式系数

poly = np.poly1d(coef)#返回多项式

（4）MLR

from numpy import genfromtxt

from sklearn import linear_model

import csv 

data = open(path,'r')
data = csv.reader(data,delimiter) #打开CSV文件格式

data = genfromtxt(filepath,delimiter)#返回矩阵

regr = linear_model.LinearRegression()

print (regr.coef_,regr.intercept_)#打印系数和截距

（5）神经网络

交叉验证法计算准确度

from sklearn.cross_validation import train_test_split
x_train,x_test,y_train,y_test = train_test_split(x,y)

有待更新。。。

转载于:https://my.oschina.net/u/3786144/blog/1813852