xMat = mat(xMat, dtype = float)
yMat = mat(yMat, dtype = float)
#Regularize the matrix
xMat[:, 1:], xMeans, xStd = regression.regularize(xMat[:, 1:])
xCheckMat[:, 1:] = (xCheckMat[:, 1:] - xMeans)/xStd
preDictPerComponent = []
for yIdx in range(0, 2): #caculate computing result and then communication result
yMatTmp = yMat[:, yIdx]
taskPredict = 0
for kElement in bestKList:
taskPredict = 0
#get the result for cared data
wr = regression.lwlr(xCheckMat, xMat,yMatTmp.T,kElement)
#wr = regression.standRegres(xMat,yMatTmp.T)
if (wr != None):
#print wr
taskPredict = xCheckMat * wr
#print taskPredict
#print "Task Size: %d, Predict Hours: %d Result[%d] is %f with k %f" % (tTSize, predictHours, yIdx, taskPredict, kElement)
#print wr
break
if taskPredict != 0:
tmp = taskPredict.reshape(-1).tolist() #sum only used for get the value but not the matrix
tmp = [j for i in tmp for j in i]
preDictPerComponent.extend(tmp)
else:
print "Error! Task Size: %d, Predict Hours: %d Task %d can not get predict value" % (tTSize, predictHours, yIdx)
fr = open("/home/yu/workspace/machine-learning-class/ex1/ex1data2.txt")
xMat = []
yMat = []
for line in fr.readlines():
tmpEle = line.split(',')
xMat.append([1, tmpEle[0], tmpEle[1]])
yMat.append([tmpEle[2]])
bestKList = [
10
] #[0.07, 0.3, 0.1, 0.7, 3, 10, 28, 40, 60, 80, 100]#[100, 80, 60, 40, 28, 10, 3, 0.7, 0.3, 0.07, 0.01]
xCheckMat = mat([1, 1650, 3], dtype=float)
xMat = mat(xMat, dtype=float)
yMat = mat(yMat, dtype=float)
#Regularize the matrix
xMat[:, 1:], xMeans, xStd = regression.regularize(xMat[:, 1:])
xCheckMat[:, 1:] = (xCheckMat[:, 1:] - xMeans) / xStd
for kElement in bestKList:
taskPredict = 0
#get the result for cared data
wr = regression.lwlr(xCheckMat, xMat, yMat.T, kElement)
#wr = regression.standRegres(xMat,yMat.T)
if (wr != None):
print wr
taskPredict = xCheckMat * wr
print taskPredict
break
# # plt.show()
#
# #相关系数
# #print('corrcoef:');print(corrcoef(yHat.T,yMat)) #预测值和真实值得匹配程度
# #corrcoef:
# # [[ 1. 0.13653777]
# # [ 0.13653777 1. ]]
# #end：Liner Regression
xArr,yArr = regression.loadDataSet('ex0.txt')
#对单点进行估计,输出预测值
print('xArr[0]:');print(xArr[0])
# xArr[0]:
# [1.0, 0.067732]
print(yArr[0]) #output: 3.176513
print(regression.lwlr(xArr[0],xArr,yArr,1.0)) #output: martix([[ 3.12204471]])
print(regression.lwlr(xArr[0],xArr,yArr,0.001)) #output: martix([[ 3.20175729]])
#为了得到数据集里所有点的估计，可以调用LwlrTest()函数：
yHat = regression.lwlrTest(xArr,xArr,yArr,0.003)
print('all points about yHat:'); print(yHat)
#查看拟合效果
xMat = mat(xArr) #xArr是什么？
srtInd = xMat[:,1].argsort(0) #对xArr排序
xSort = xMat[srtInd][:,0,:]
#绘图：
fig = plt.figure()
ax = fig.add_subplot(111)
ax.plot(xSort[:,1],yHat[srtInd])
ax.scatter(xMat[:,1].flatten().A[0],mat(yArr).T.flatten().A[0],s=2,c='red')
yHat = xMat * ws #绘出数据集散点图和最佳拟合直线图 import matplotlib.pyplot as plt fig = plt.figure() ax = fig.add_subplot(111) ax.scatter(xMat[:, 1].flatten().A[0], yMat.T[:, 0].flatten().A[0]) #为了绘制计算出的最佳拟合曲线，需要绘出yHat的值 #若直线上的数据点次序混乱，绘图时将会出现问题，固要先将点按照升序排列 xCopy = xMat.copy() xCopy.sort(0) #这个应该是np中的sort，意思是按照0维度排序 yHat = xCopy * ws ax.plot(xCopy[:, 1], yHat) plt.show() #对单点进行估计 print yArr[0] print regression.lwlr(xArr[0], xArr, yArr, 1.0) print regression.lwlr(xArr[0], xArr, yArr, 0.001) #得到所有点的估计 yHat = regression.lwlrTest(xArr, xArr, yArr, 0.003) srtInd = xMat[:, 1].argsort(0) xSort = xMat[srtInd][:, 0, :] fig = plt.figure() ax = fig.add_subplot(111) ax.plot(xSort[:, 1], yHat[srtInd]) ax.scatter(xMat[:, 1].flatten().A[0], mat(yArr).T.flatten().A[0], s=2, c='red') plt.show()
ax = fig.add_subplot(111)
ax.scatter(xMat[:, 1].flatten().A[0], yMat.T[:, 0].flatten().A[0])
xCopy = xMat.copy()
xCopy.sort(0)
yHat = xCopy * ws
ax.plot(xCopy[:, 1], yHat)
plt.show()
# 判断模型效果
yHat = xMat * ws
corrcoef(yHat.T, yMat) # 看[0,1]和[1,0]位置，因为自己和自己相关性肯定是1
# 测试局部加权线性回归
reload(regression)
xArr, yArr = regression.loadDataSet('ex0.txt')
regression.lwlr(xArr[0], xArr, yArr, 1.0) # 单点测试
regression.lwlr(xArr[0], xArr, yArr, 0.001)
yHat = regression.lwlrTest(xArr, xArr, yArr, 0.5) # 预测xArr,可以给不同的k测试不同的效果
# 绘制估计值和原始值
xMat = mat(xArr)
srtInd = xMat[:, 1].argsort(0)
xSort = xMat[srtInd][:, 0, :]
fig = plt.figure()
ax = fig.add_subplot(111)
ax.plot(xSort[:, 1], yHat[srtInd])
ax.scatter(xMat[:, 1].flatten().A[0], mat(yArr).T.flatten().A[0], s=2, c='red')
plt.show()
# 在真实数据上
reload(regression)
ax = fig.add_subplot(111)
ax.scatter(xMat[:, 1].flatten().A[0], yMat.T[:, 0].flatten().A[0])
xCopy = xMat.copy()
xCopy.sort(0)
yHat = xCopy * ws
ax.plot(xCopy[:, 1], yHat)
plt.show()
yHat = xMat * ws
print "corrcoef(yHat.T,yMat):", corrcoef(yHat.T, yMat)
#8.2 局部加权线性回归
xArr, yArr = regression.loadDataSet(homedir + 'ex0.txt')
print "yArr[0]:", yArr[0]
print "xArr[0]:", xArr[0]
print "regression.lwlr(xArr[0],xArr,yArr,1.0):", regression.lwlr(
xArr[0], xArr, yArr, 1.0)
print "regression.lwlr(xArr[0],xArr,yArr,0.001):", regression.lwlr(
xArr[0], xArr, yArr, 0.001)
print ":",
yHat = regression.lwlrTest(xArr, xArr, yArr, 0.02)
xMat = mat(xArr)
srtInd = xMat[:, 1].argsort(0)
xSort = xMat[srtInd][:, 0, :]
# print "xMat",xMat
# print "srtInd:",srtInd
# print "xSort:",xSort
fig = plt.figure()
ax = fig.add_subplot(111)
ax.plot(xSort[:, 1], yHat[srtInd])
ax.scatter(xMat[:, 1].flatten().A[0], mat(yArr).T.flatten().A[0], s=2, c='red')
plt.show()
#
# import matplotlib.pyplot as plt
#
# fig=plt.figure()
# ax=fig.add_subplot(111)
#
# ax.scatter(xMat[:,1].flatten().A[0],yMat.T[:,0].flatten().A[0])
#
# xCopy=xMat.copy();
# xCopy.sort(0)
# yHat=xCopy*ws
#
# ax.plot(xCopy[:,1],yHat)
# plt.show()
ws = regression.lwlr(xArr[0], xArr, yArr, 1.0)
print(ws)
yHat = regression.lwlrTest(xArr, xArr, yArr, 0.03)
xMat = mat(xArr)
srtInd = xMat[:, 1].argsort(0)
xSort = xMat[srtInd][:, 0, :]
import matplotlib.pyplot as plt
fig = plt.figure()
ax = fig.add_subplot(111)
ax.scatter(xMat[:, 1].flatten().A[0],
mat(yArr).T[:, 0].flatten().A[0],
s=2,
fig = plt.figure() #创建画布
ax = fig.add_subplot(211)
#ax = fig.add_subplot(349) 参数349的意思是：将画布分割成3行4列，图像画在从左到右从上到下的第9块，
# 3410是不行的，可以用另一种方式(3,4,10)。
# ax = fig.add_subplot(2,1,1)
# ax.plot(x,y)
# ax = fig.add_subplot(2,2,3)
# ax.plot(x,y)
# plt.show() 显示多图
ax.scatter(xMat[:, 1].flatten().A[0], yMat.T[:, 0].flatten().A[0])
xCopy = xMat.copy()
xCopy.sort(0)
yHat = xCopy * ws
ax.plot(xCopy[:, 1], yHat)
print corrcoef(yHat.T, yMat) # 转置 保证都是行向量
xArrL, yArrL = regression.loadDataSet('ex0.txt')
yArrL[0]
regression.lwlr(xArrL[0], xArrL, yArrL, 1.0)
yHatL = regression.lwlrTest(xArrL, xArrL, yArrL, 0.01)
xMatL = mat(xArrL)
yMatL = mat(yArrL)
srtInd = xMatL[:, 1].argsort(0)
xSort = xMatL[srtInd][:, 0, :]
ax = fig.add_subplot(2, 1, 2)
ax.scatter(xMatL[:, 1].flatten().A[0], yMatL.T[:, 0].flatten().A[0])
ax.plot(xSort[:, 1], yHatL[srtInd])
plt.show()
print corrcoef(yHatL.T, yMatL)
import regression xArr, yArr = regression.loadDataSet() # print(xArr[0:2]) w = regression.standRegres(xArr, yArr) # print(w) print(yArr[0], regression.lwlr(xArr[0], xArr, yArr, 0.001)) yHat = regression.lwlrTest(xArr, xArr, yArr, 0.003) print(yHat) xMat = np.mat(xArr) srtInd = xMat[:, 1].argsort(0) #? xSort = xMat[srtInd][:, 0, :] #? fig = plt.figure() ax = fig.add_subplot(111) ax.plot(xSort[:, 1], yHat[srtInd]) ax.scatter(xMat[:, 1].flatten().A[0], mat(yArr).T.flatten().A[0], s=2, c='red') plt.show()
xCopy.sort(0)
yHat = xCopy * ws
ax.plot(xCopy[:, 1], yHat)
plt.show()
xMat = mat(dm)
yMat = mat(ls)
yHat = xMat * ws
corrcoef(yHat.T, yMat) # 求相关系数
# 局部加权线性回归
import regression
from numpy import *
dm, ls = regression.loadDataSet('ex0.txt')
ls[0]
regression.lwlr(dm[0], dm, ls, 1.0)
yHat = regression.lwlrTest(dm, dm, ls, 0.01)
xMat = mat(dm)
strInd = xMat[:, 1].argsort(0)
xSort = xMat[strInd][:, 0, :]
import matplotlib.pyplot as plt
fig = plt.figure()
ax = fig.add_subplot(111)
ax.plot(xSort[:, 1], yHat[strInd])
ax.scatter(xMat[:, 1].flatten().A[0], mat(ls).T.flatten().A[0], s=2, c='red')
plt.show()
# 预测鲍鱼寿命
import regression
from numpy import *
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
'8.2'
__author__ = 'lxp'
import regression
import numpy as np
import matplotlib.pyplot as plt
xArr, yArr = regression.loadDataSet('ex0.txt')
print(regression.lwlr(xArr[0], xArr, yArr, 1))
print(regression.lwlr(xArr[0], xArr, yArr, 0.001))
yHat = regression.lwlrTest(xArr, xArr, yArr, 0.003)
xMat = np.mat(xArr)
srtInd = xMat[:, 1].argsort(0)
xSort = xMat[srtInd][:, 0, :]
fig = plt.figure()
ax = fig.add_subplot(111)
ax.plot(xSort[:, 1], yHat[srtInd])
ax.scatter(xMat[:, 1].flatten().A[0],
np.mat(yArr).T.flatten().A[0],
s=2,
c='red')
plt.show()
xArr_origin, yArr_origin = regression.loadDataSet("foo.txt")
# print(yArr_origin)
xArr = []
yArr = []
m = (shape(xArr_origin))[0]
yHat = zeros(m)
xArr.append(xArr_origin[0][:])
yArr.append(yArr_origin[0])
yHat[0] = yArr_origin[0]
for i in range(1, m):
# k = best_k(xArr, yArr)
k = 0.06830078125
x = xArr_origin[i][:]
y = regression.lwlr(x, xArr, yArr, k)
# print(y.flatten().A[0][0])
yHat[i] = y.flatten().A[0][0]
while yHat[i] <= 0:
k = k + 0.01
y = regression.lwlr(x, xArr, yArr, k)
yHat[i] = y.flatten().A[0][0]
xArr.append(x)
yArr.append(yArr_origin[i])
print(i, k, yHat[i], yArr[i])
with open('workfile', 'a') as f:
f.write(str(i))
f.write("\t")
f.write(str(k))
f.write("\t")
f.write(str(yHat[i]))
xMat, yMat = mat(xArr), mat(yArr)
yHat = xMat * ws
import matplotlib.pyplot as plt
fig = plt.figure()
ax = fig.add_subplot(111)
ax.scatter(xMat[:,1].flatten().A[0], yMat.T[:,0].flatten().A[0])
xCopy = xMat.copy()
xCopy.sort(0)
yHat = xCopy * ws
ax.plot(xCopy[:,1], yHat)
plt.show()
yHat = xMat * ws
print corrcoef(yHat.T, yMat)
xArr, yArr = regression.loadDataSet('ex0.txt')
print yArr[0]
print regression.lwlr(xArr[0], xArr, yArr, 1.0)
print regression.lwlr(xArr[0], xArr, yArr, 0.001)
yHat = regression.lwlrTest(xArr, xArr, yArr, 0.01)
# 1.0 与标准回归一致 欠拟合
# 0.003 过拟合
xMat = mat(xArr)
srtInd = xMat[:,1].argsort(0)
xSort = xMat[srtInd][:,0,:]
#import matplotlib.pyplot as plt
fig = plt.figure()
ax = fig.add_subplot(111)
ax.plot(xSort[:,1], yHat[srtInd])
ax.scatter(xMat[:,1].flatten().A[0], mat(yArr).T.flatten().A[0], s=2, c='red')
plt.show()
# wx = reg.standRegres(xArr, yArr) # print wx # # xMat = mat(xArr) # yMat = mat(yArr) # # yHat = xMat * wx # corrcoef(yHat.H, yMat) # # fig = plt.figure() # ax = fig.add_subplot(111) # ax.scatter(xMat[:, 1].flatten().A[0], yMat.T[:, 0].flatten().A[0]) # xCopy = xMat.copy() # xCopy.sort(0) # yHat = xCopy * wx # ax.plot(xCopy[:, 1], yHat) # plt.show() print reg.lwlr(xArr[0], xArr, yArr, 1.0) print reg.lwlr(xArr[0], xArr, yArr, 0.001) yHat = reg.lwlrTest(xArr, xArr, yArr, 0.003) xMat = mat(xArr) srtInd = xMat[:, 1].argsort(0) xSort = xMat[srtInd][:, 0, :] fig = plt.figure() ax = fig.add_subplot(111) ax.plot(xSort[:, 1], yHat[srtInd]) ax.scatter(xMat[:, 1].flatten().A[0], mat(yArr).T.flatten().A[0], s=2, c='red') plt.show();
