IDE:PyCharm Edu 4.02
環境:Adaconda3 python3.6
關鍵詞:sigmoid函數、批梯度上升法、隨機梯度上升法
from numpy import *
import matplotlib.pyplot as plt
def loadDataSet():
dataMat = []
labelMat = []
with open('testSet.txt') as fr:
for line in fr.readlines():
lineArr = line.strip().split()
dataMat.append([1.0,float(lineArr[0]),float(lineArr[1])])
labelMat.append(int(lineArr[2]))
return dataMat,labelMat
dataMat,labelMat = loadDataSet()
def sigmoid(inX):
return 1.0/(1+exp(-inX))
# 批梯度上升算法(計算量大)
def gradAscent(dataMatIn,classLabels):
#convert to NumPy matrix
dataMatrix = mat(dataMatIn) # 100 by 3
labelMat = mat(classLabels).transpose() # 100 by 1
m,n = shape(dataMatrix)
alpha = 0.001
maxCycles = 500 #迭代次數
weights = ones((n,1)) #矩陣 3 by 1
for k in range(maxCycles):
h = sigmoid(dataMatrix*weights) # 兩個矩陣類型 *表示矩陣乘法
error = labelMat-h
weights = weights + alpha * dataMatrix.transpose() * error #批梯度下降法公式
return weights
weights1 = gradAscent(dataMat,labelMat)
#print(weights1) # print(weights1.getA())
# 畫出數據集和logistic迴歸最佳擬合直線
def plotBestFit(weights):
dataArr = array(dataMat) #二維時,array()與mat()函數效果相同
n = shape(dataArr)[0] #行數
xcord1 = [];ycord1 = []
xcord2 = [];ycord2 = []
for i in range(n):
if int(labelMat[i]) == 1:
xcord1.append(dataArr[i,1])
ycord1.append(dataArr[i,2])
else:
xcord2.append(dataArr[i,1])
ycord2.append(dataArr[i,2])
fig = plt.figure()
ax = fig.add_subplot(111)
ax.scatter(xcord1,ycord1,s=30,c='red',marker='s')
ax.scatter(xcord2,ycord2,s=30,c='green')
x = arange(-3.0,3.0,0.1) # 數組(60,)
#根據sigmoid(z)函數,0是兩個分類的分界處
#z=w0x0+w1x1+w2x2 令z=0,x0=1解出x1與x2的關係
y = (-weights[0]-weights[1]*x)/weights[2] # 矩陣(1,60)
# 原文 ax.plot(x,y)
ax.plot(x,y.transpose())
plt.xlabel('X1');plt.ylabel('X2')
plt.show()
#原文 plotBestFit(weights1.getA())
#print(plotBestFit(weights1))
# 隨機梯度上升算法
def stocGradAscent0(dataMatrix,classLabels):
m,n = shape(dataMatrix)
alpha = 0.01
weights = ones(n) # 數組
for i in range(m):
h = sigmoid(sum(dataMatrix[i]*weights)) # 元素相乘再求和即w0x0+w1x1+w2x2
error = classLabels[i] - h
weights = weights + alpha * error * dataMatrix[i]
return weights
weights2 = stocGradAscent0(array(dataMat),labelMat)
# print(weights2)
# print(plotBestFit(weights2))
# 改進的隨機梯度下降法
# alpha隨着迭代次數不斷減小
def stocGradAscent1(dataMatrix,classLabels,numIter=150):
m,n = shape(dataMatrix)
weights = ones(n) # 數組對象
dataMatrix = array(dataMatrix) #轉換爲numpy格式
for j in range(numIter):
# 原文 dataIndex = range(m)
dataIndex = list(range(m))
for i in range(m):
# 隨機選擇一個樣本進行權重的更新
alpha = 4/(1.0+j+i)+0.001 #apha decreases with iteration, does not
randIndex = int(random.uniform(0,len(dataIndex))) #go to 0 because of the constant
h = sigmoid(sum(dataMatrix[randIndex]*weights))
error = classLabels[randIndex] - h
weights = weights + alpha * error * dataMatrix[randIndex]
del(dataIndex[randIndex])
return weights
weights3 = stocGradAscent1(dataMat,labelMat)
print(plotBestFit(weights3))
註解:
1、numpy:矩陣和數組的轉換
np.mat(變量)函數 :將對象轉換爲matrix
np.變量.getA():將矩陣轉換爲數組
例子:批梯度下降法返回一個矩陣weight1s,而plotBestFit(weights)函數接收一個數組,
因此,調用命令爲plotBestFit(weights1.getA())。
直接使用 plotBestFit(weights1)報錯:x and y must have same first dimension, but have shapes (60,) and (1, 60)
解決方法:將原文的ax.plot(x,y) 改爲ax.plot(x,y.transpose())
2、區分list、numpy的矩陣及數組
(1)
list對象中間有逗號[1,1,1,1,1]
print(ones(5)) #數組
print(ones((5,1))) #矩陣
[ 1. 1. 1. 1. 1.]
[[ 1.]
[ 1.]
[ 1.]
[ 1.]
[ 1.]]
(2)
矩陣對象:* 表示矩陣乘法
ndarray對象: * 表示元素乘法;dot(A,B)表示矩陣乘法。
當然,二維的ndarray與matrix相同。
python列表:print([1,2,3]*2) 結果:[1,2,3,1,2,3]
若想從列表得到數乘結果,可以使用列表生成式!
(3)記住隨機梯度上升法的推導公式
weights = weights + alpha * error * dataMatrix[randIndex]