feature, value = ChooseBestFeature(dataArray, resultList, maxDataCount,
minErrorReduce)
if feature == None:
#Leaf
node = Node(0, value, None, None)
else:
subDataArray1, subResultList1, subDataArray2, subResultList2 = SplitDataset(
dataArray, resultList, feature, value)
leftTree = CreateModelTree(subDataArray1, subResultList1, maxDataCount,
minErrorReduce)
rightTree = CreateModelTree(subDataArray2, subResultList2,
maxDataCount, minErrorReduce)
node = Node(feature, value, leftTree, rightTree)
return node
def ForecastByModelTree(tree, inData):
if (tree.left == None and tree.right == None):
return (numpy.mat(inData) * tree.value).tolist()[0][0]
if (inData[tree.feature] > tree.value):
return ForecastByModelTree(tree.left, inData)
else:
return ForecastByModelTree(tree.right, inData)
if __name__ == '__main__':
dataArray, resultList = LoadDataAndLabel("Dataset2.txt")
tree = CreateModelTree(dataArray, resultList, 4, 0.1)
#ShowTree(tree)
print(ForecastByModelTree(tree, [0.530897]))
print(ForecastByModelTree(tree, [0.993349]))
#Caluclate the b in two conditions
b1 = b - Ei - labelMatrix[i] * (
newAlphaI - alphas[i]) * dataMatrix[
i, :] * dataMatrix[i, :].T - labelMatrix[j] * (
newAlphaJ - alphas[j]
) * dataMatrix[i, :] * dataMatrix[j, :].T
b2 = b - Ej - labelMatrix[i] * (
newAlphaI - alphas[i]) * dataMatrix[
i, :] * dataMatrix[j, :].T - labelMatrix[j] * (
newAlphaJ - alphas[j]
) * dataMatrix[j, :] * dataMatrix[j, :].T
#Update the alpha[i], alpha[j] and b
alphas[i] = newAlphaI
alphas[j] = newAlphaJ
if (0 < alphas[i] and alphas[i] < border):
b = b1
elif (0 < alphas[j] and alphas[j] < border):
b = b2
else:
b = (b1 + b2) / 2.0
iter = iter + 1
return alphas, b
if __name__ == '__main__':
dataArray, labelList = LoadDataAndLabel("Dataset.txt")
alphas, b = smoSimple(dataArray, labelList, 0.6, 0.001, 100)
print(alphas)
print(b)
errorCount += 1
print("The error rate:%f" % (errorCount / float(numToTest)))
#绘图
def Plot(dataMatrix, labelList):
fig = plt.figure()
ax = fig.add_subplot(111)
ax.scatter(dataMatrix[:, 1],dataMatrix[:, 2])
ax.scatter(dataMatrix[:, 1],dataMatrix[:, 2], 15.0 * array(labelList), 15.0 * array(labelList))
plt.show()
def ClassifyPerson(normMatrix, ranges, minValues, labelList):
resultList = ['Not in all','In small doses','In large doses']
percentTats = float(input("Percentage of time spent playing video game?"))
ffMiles = float(input("Frequent flier miles earned per year?"))
iceCream = float(input("Liters of ice cream consumed per year?"))
inArray = [ffMiles,percentTats,iceCream]
classifierResult = Classify0((inArray)/ranges,normMatrix,labelList,3)
print("You will probably like this person:",resultList[int(classifierResult)-1])
if __name__ == '__main__':
dataArray,labelList = LoadDataAndLabel('DatingTestSet2.txt')
dataMatrix = array(dataArray)
normMatrix,ranges,minValues = AutoNorm(dataMatrix)
DataClassTest(normMatrix, ranges, minValues, labelList)
ClassifyPerson(normMatrix, ranges, minValues, labelList)
Plot(dataMatrix, labelList)
x1 = min(X)
y1 = w[0] + x1 * w[1]
x2 = max(X)
y2 = w[0] + x2 * w[1]
ax.plot([x1, x2], [y1, y2], 'r-')
sortedXIndex = numpy.argsort(X)
for index in range(len(X) - 1):
ax.plot([X[sortedXIndex[index]], X[sortedXIndex[index + 1]]], [
LWLRPredY[sortedXIndex[index]], LWLRPredY[sortedXIndex[index + 1]]
], 'g--')
matplotlib.pyplot.show()
if __name__ == '__main__':
dataArray, labelList = LoadDataAndLabel('Dataset1.txt')
w = StandRegres(dataArray, labelList)
StandPredY = [numpy.mat(dataArray[i]) * w for i in range(len(dataArray))]
LWLRPredY = [
LocalWeightLinearRegress(data, dataArray, labelList,
0.01).tolist()[0][0] for data in dataArray
]
w1 = RidgeRegres(dataArray, labelList)
RidgePredY = [(w1[0] + dataArray[i][1] * w1[1]).tolist()[0][0]
for i in range(len(dataArray))]
x = [i[1] for i in dataArray]
Plot(x, labelList, w.flatten().A[0], LWLRPredY)
print(RegError(labelList, StandPredY))
print(RegError(labelList, RidgePredY))