写在开头的话：在学习《机器学习实战》的过程中发现书中很多代码并没有注释，这对新入门的同学是一个挑战，特此贴出我对代码做出的注释，仅供参考，欢迎指正。
1、将CART算法用于回归
#coding:gbk
from numpy import *
#作用：从文件导入数据
#输入：文件名
#输出：数据矩阵
def loadDataSet(fileName):
dataMat = []
fr = open(fileName)
for line in fr.readlines():
curLine = line. strip().split('\t')
fltLine = map(float, curLine)#将每行映射成浮点数
dataMat.append(fltLine)
return dataMat
# 作用：从文件导入数据
# 输入：数据矩阵，待切分特征值，阈值
# 输出：切分后的数据集
def binSplitDataSet(dataSet, feature, value):
# 书中最后有[0]，练习发现只会返回1*n矩阵，因此删掉
mat0 = dataSet[nonzero(dataSet[:, feature] > value)[0], :]# nonzero()返回的是列表的下标值
mat1 = dataSet[nonzero(dataSet[:, feature] <= value)[0], :]
return mat0, mat1
# 作用：目标变量的均值
# 输入：数据集
# 输出：目标变量的均值
def regLeaf(dataSet):
return mean(dataSet[:, -1])
# 作用：目标变量的总方差
# 输入：数据集
# 输出：目标变量的总方差
def regErr(dataSet):
return var(dataSet[:, -1]) * shape(dataSet)[0]#均方差乘以数据集中样本的个数，得总方差
# 作用：找到数据的最佳二元切分方式
# 输入：数据集，建立叶节点的函数，误差计算函数，包含树构建所需其他参数的元组
# 输出：最佳切分特征和特征值
def chooseBestSplit(dataSet, leafType = regLeaf, errType = regErr, ops = (1, 4)):
tolS = ops[0]#容许的误差下降值
tolN = ops[1]#切分的最少样本数
#如果所有值都相等则退出
if len(set(dataSet[:, -1].T.tolist()[0])) == 1:
return None, leafType(dataSet)
m, n = shape(dataSet)#数据集的行数和列数
S = errType(dataSet)#数据集的总方差
bestS = inf
bestIndex = 0
bestValue = 0
for featIndex in range(n-1):#最后一列为y值
#书中代码有错，需改成如下形式，转置后转换为列表
for splitVal in set(dataSet[:, featIndex].T.tolist()[0]):#对每个特征的特征值
mat0, mat1 = binSplitDataSet(dataSet, featIndex, splitVal)
# 有切分的数据集太小，跳过该种切分方式
if (shape(mat0)[0] < tolN) or (shape(mat1)[0] < tolN):
continue
#切分后的总方差
newS = errType(mat0) + errType(mat1)
if newS < bestS:
bestIndex = featIndex
bestValue = splitVal
bestS = newS
#如果误差不大则退出
if (S - bestS) < tolS:
return None, leafType(dataSet)
mat0, mat1 = binSplitDataSet(dataSet, bestIndex, bestValue)
#如果切分出的数据集很小则退出
if (shape(mat0)[0] < tolN) or (shape(mat1)[0] < tolN):
return None, leafType(dataSet)
return bestIndex, bestValue
#作用：创建树
#输入：数据集，建立叶节点的函数，误差计算函数，包含树构建所需其他参数的元组
#输出：树
def createTree(dataSet, leafType = regLeaf, errType = regErr, ops = (1, 4)):
feat, val = chooseBestSplit(dataSet, leafType, errType, ops)
if feat == None:
return val
retTree = {}
retTree['spInd'] = feat
retTree['spVal'] = val
lSet, rSet = binSplitDataSet(dataSet, feat, val)
retTree['left'] = createTree(lSet, leafType, errType, ops)
retTree['right'] = createTree(rSet, leafType, errType, ops)
return retTree
#作用：测试输入变量是否是一棵树
#输入：输入变量
#输出：布尔类型的结果，是一棵树则返回True
def isTree(obj):
return (type(obj).__name__ == 'dict')#如果是一棵树，则类型为'dict'即字典
#作用：得到树的平均值
#输入：树
#输出：数的平均值
def getMean(tree):
if isTree(tree['right']):
tree['right'] = getMean(tree['right'])
if isTree(tree['left']):
tree['left'] = getMean(tree['left'])
#如果左子树和右子树的样例个数不相等，意义是什么？
return (tree['left'] + tree['right']) / 2.0
#作用：剪枝
#输入：待剪枝的树，剪枝所需的测试数据
#输出：剪好的树
def prune(tree, testData):
if shape(testData)[0] == 0:#确认测试集是否为空
return getMean(tree)
#如果两个分支有一个是子树，则对测试数据进行划分
if (isTree(tree['right']) or isTree(tree['left'])):
lSet, rSet = binSplitDataSet(testData, tree['spInd'], tree['spVal'])
# 对左子树进行剪枝
if isTree(tree['left']):
tree['left'] = prune(tree['left'], lSet)
# 对右子树进行剪枝
if isTree(tree['right']):
tree['right'] = prune(tree['right'], rSet)
#两个分支都不是子树
if not isTree(tree['left']) and not isTree(tree['right']):
lSet, rSet = binSplitDataSet(testData, tree['spInd'], tree['spVal'])
#没有合并的误差
errorNoMerge = sum(power(lSet[:, -1] - tree['left'], 2)) + \
sum(power(rSet[:, -1] - tree['right'], 2))
#有点疑问，如果左右子树样例个数不相等，则treeMean意义是什么？
treeMean = (tree['left'] + tree['right']) / 2.0
#合并后的误差
errorMerge = sum(power(testData[:, -1] - treeMean, 2))
if errorMerge < errorNoMerge:
print "merging"
return treeMean
else:
return tree
else:
return tree2、模型树
#作用：将数据集格式化成目标变量Y和自变量X，并求回归系数向量
#输入：数据集
#输出：回归系数向量，自变量，目标变量
def linearSolve(dataSet):
m, n = shape(dataSet)
X = mat(ones((m, n)))
Y = mat(ones((m, 1)))
X[:, 1:n] = dataSet[:, 0:n-1]#X[:, 0]为截距，均为1
Y = dataSet[:, -1]
xTx = X.T * X
if linalg.det(xTx) == 0.0:
raise NameError('This matrix is singular, cannot do inverse,\n\
try increasing the second value of ops')
ws = xTx.I * (X.T * Y)
return ws, X, Y
#作用：求数据集的回归系数向量
#输入：数据集
#输出：回归系数向量
def modelLeaf(dataSet):
ws, X, Y = linearSolve(dataSet)
return ws
#作用：求数据集的方差
#输入：数据集
#输出：数据集的方差
def modelErr(dataSet):
ws, X, Y = linearSolve(dataSet)
yHat = X * ws
return sum(power(Y - yHat, 2))
#作用：返回模型的浮点数
#输入：模型，？
#输出：模型的浮点数
def regTreeEval(model, inDat):
return float(model)
#作用：貌似没有出现过
#输入：
#输出：
def modelTreeEval(model, inDat):
n = shape(inDat)[1]
X = mat(ones((1, n + 1)))
X[:, 1:n+1] = inDat
return float(X * model)
#作用：预测测试数据在树中的值
#输入：数，测试数据，树形式
#输出：测试数据在树中的值
def treeForeCast(tree, inData, modelEval = regTreeEval):
if not isTree(tree):
return modelEval(tree, inData)
if inData[tree['spInd']] > tree['spVal']:#inData[]值比根节点大，进入左子树
if isTree(tree['left']):
return treeForeCast(tree['left'], inData, modelEval)
else:
return modelEval(tree['left'], inData)
else:#inData[]值比根节点小，进入右子树
if isTree(tree['right']):
return treeForeCast(tree['right'], inData, modelEval)
else:
return modelEval(tree['right'], inData)
#作用：预测测试数据集在树中的值
#输入：数，测试数据集，树形式
#输出：测试数据集在树中的值
def createForeCast(tree, testData, modelEval = regTreeEval):
m = len(testData)
yHat = mat(zeros((m, 1)))
for i in range(m):
yHat[i, 0] = treeForeCast(tree, mat(testData[i]), modelEval)
return yHat3、使用Python的Tkinter库创建GUI
#coding:gbk
from numpy import *
from Tkinter import *
import regTrees
import matplotlib
matplotlib.use('TkAgg')
from matplotlib.backends.backend_tkagg import FigureCanvasTkAgg
from matplotlib.figure import Figure
#作用：重绘新面板
#输入：容许的误差下降值，切分的最少样本数
#输出：无
def reDraw(tolS, tolN):
reDraw.f.clf()
reDraw.a = reDraw.f.add_subplot(111)
#检查复选框是否选中
if chkBtnVar.get():#复选框被选中，执行模型树
if tolN < 2:#切分的最少样本数不能少于2
tolN = 2
myTree = regTrees.createTree(reDraw.rawDat, regTrees.modelLeaf, regTrees.modelErr, (tolS, tolN))
yHat = regTrees.createForeCast(myTree, reDraw.testDat, regTrees.modelTreeEval)
else:#复选框没被选中，执行会归树
myTree = regTrees.createTree(reDraw.rawDat, ops = (tolS, tolN))
yHat = regTrees.createForeCast(myTree, reDraw.testDat)
#绘制sine.txt中的点
reDraw.a.scatter(reDraw.rawDat[:, 0], reDraw.rawDat[:, 1], s = 5)
#绘制得到的直线
reDraw.a.plot(reDraw.testDat, yHat, linewidth = 2.0)
reDraw.canvas.show()
#作用：得到tolN和tolS的取值
#输入：无
#输出：无
def getInputs():
# 得到tolN的取值
try:
tolN = int(tolNentry.get())
except:
tolN = 10
print "enter Integer for tolN"
tolNentry.delete(0, END)
tolNentry.insert(0, '10')
# 得到tolS的取值
try:
tolS = float(tolSentry.get())
except:
tolS = 1.0
print "enter Float for tolS"
tolSentry.delete(0, END)
tolSentry.insert(0, '1.0')
return tolN, tolS
#作用：绘制新的树
#输入：无
#输出：无
def drawNewTree():
tolN, tolS = getInputs()
reDraw(tolS, tolN)
root = Tk()
#绘制主面板
reDraw.f = Figure(figsize = (5, 4), dpi = 100)
reDraw.canvas = FigureCanvasTkAgg(reDraw.f, master = root)
reDraw.canvas.show()
reDraw.canvas.get_tk_widget().grid(row = 0, columnspan = 3)
#创建退出按钮
#Button(root, text = "Quit", fg = 'black', command = root.quit).grid(row = 1, column = 2)
#插入标题标签
#Label(root, text = "Plot Place Holder").grid(row = 0, columnspan = 3)
#插入tolN标签
Label(root, text = "tolN").grid(row = 1, column = 0)
#插入tolN文本框
tolNentry = Entry(root)
tolNentry.grid(row = 1, column = 1)
tolNentry.insert(0, '10')
#插入tolS标签
Label(root, text = "tolS").grid(row = 2, column = 0)
#插入tolS文本框
tolSentry = Entry(root)
tolSentry.grid(row = 2, column = 1)
tolSentry.insert(0, '1.0')
#创建ReDraw按钮
Button(root, text = "ReDraw", command = drawNewTree).grid(row = 1, column = 2, rowspan = 3)
#创建按钮勾选状态变量
chkBtnVar = IntVar()
#创建Model Tree按钮
chkBtn = Checkbutton(root, text = "Model Tree", variable = chkBtnVar)
chkBtn.grid(row = 3, column = 0, columnspan = 2)
#sine.txt中的数据集
reDraw.rawDat = mat(regTrees.loadDataSet('exp2.txt'))
#sine.txt第一行即x轴的最大值到最小值
reDraw.testDat = arange(min(reDraw.rawDat[:, 0]), max(reDraw.rawDat[:, 0]), 0.01)
reDraw(1.0, 10)
root.mainloop()
扫一扫
5万+
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
