def standRegres(page_name,hotnum,timeline,percent = '',notshow = None):
from fan import sharefan
from fan import likesum_nothotfan
from numpy import *
#loaddata
dataMat = []
labelMat = []
(hotfan,onehoursdict,msgdict) = sharefan(page_name=page_name,hotfannum=hotnum,hours=timeline,percent=percent,justgetdata=True)
timedict = likesum_nothotfan(page_name=page_name,hotnum=hotnum,timeline=timeline,percent=percent,justgetdata=True)
lendatanum = 0
for i in onehoursdict:
if i in timedict:
tempx = 0
#lendatanum += 1
if not onehoursdict[i]:
if msgdict[i][0]>50:
print str(msgdict[i][0])+msgdict[i][2]+str(msgdict[i][1])+' '+i
continue
lendatanum += 1
for j in hotfan:
if (j,) in onehoursdict[i]:
tempx += 1
x = tempx/len(onehoursdict[i])
if msgdict[i][0]:
if x <=0 :
x = 1
if timedict[i]['like_sum'] <= 0:
timedict[i]['like_sum'] = 1
y = msgdict[i][0]
dataMat.append([log10(x),log10(timedict[i]['like_sum']),1.0])
labelMat.append(log10(y))
#计算w
print '总条数：'+str(lendatanum)
print '总数据：'+str(len(dataMat))
testx = []
testy = []
(dataMat,labelMat,testx,testy) = devide_data(dataMat,labelMat)
print '训练数据：'+str(len(dataMat))
print '测试数据：'+str(len(testx))
xMat = mat(dataMat); yMat = mat(labelMat).T
xTx = xMat.T*xMat
if linalg.det(xTx) == 0.0:
print '没有解'
return
ws = xTx.I * (xMat.T*yMat)
return drawtest(dataMat,labelMat,ws.getA1(),page_name,hotnum,timeline,testx,testy,percent=percent,xlabelinput=['sharehotfan','like_not_fan'],titleinput='like_sum_notfan and sharehotfan:share_count',notshow = notshow)
def standRegres_share(page_name,hotnum,timeline,percent = '',notshow = None):
from fan import sharefan
from fan import likesum_nothotfan
from numpy import *
#loaddata
dataMat = []
labelMat = []
(hotfan,onehoursdict,msgdict) = sharefan(page_name=page_name,hotfannum=hotnum,hours=timeline,percent=percent,justgetdata=True)
id2num = dict()
lendatanum = 0
for i in onehoursdict:
if onehoursdict[i] and msgdict[i]:
id2num[i] = lendatanum
lendatanum += 1
tempx = 0
for j in onehoursdict[i]:
if j[0] in hotfan:
tempx += 1
x = tempx/len(onehoursdict[i])
if msgdict[i][0]:
y = msgdict[i][0]
if x > 0:
dataMat.append([log10(x),log10(len(onehoursdict[i])),1.0])
else:
x = 1
dataMat.append([log10(x),log10(len(onehoursdict[i])),1.0])
labelMat.append(log10(y))
xMat = mat(dataMat); yMat = mat(labelMat).T
xTx = xMat.T*xMat
if linalg.det(xTx) == 0.0:
print '没有解'
return
ws = xTx.I * (xMat.T*yMat)
ws = ws.getA1()
print ws
while(1):
test_id = raw_input('please input id:')
if test_id == 'q':
break
print xMat[id2num[test_id]]
tempx = dataMat[id2num[test_id]]
print tempx
testresult = 10**(ws[0]*tempx[0] + ws[1]*tempx[1] + ws[2])
print '预测结果:' + str(testresult)
print '真实结果:' + str(msgdict[test_id][0])
def standRegres_share(page_name,hotnum,timeline,percent = '',notshow = None):
from fan import sharefan
from fan import likesum_nothotfan
from numpy import *
#loaddata
dataMat = []
labelMat = []
(hotfan,onehoursdict,msgdict) = sharefan(page_name=page_name,hotfannum=hotnum,hours=timeline,percent=percent,justgetdata=True)
lendatanum = 0
for i in onehoursdict:
if onehoursdict[i] and msgdict[i]:
lendatanum += 1
tempx = 0
for j in onehoursdict[i]:
if j[0] in hotfan:
tempx += 1
x = tempx/len(onehoursdict[i])
if msgdict[i][0]:
y = msgdict[i][0]
if x > 0:
dataMat.append([log10(x),log10(len(onehoursdict[i])),1.0])
else:
x = 1
dataMat.append([log10(x),log10(len(onehoursdict[i])),1.0])
labelMat.append(log10(y))
print '总条数：'+str(lendatanum)
print '总数据：'+str(len(dataMat))
testx = []
testy = []
(dataMat,labelMat,testx,testy) = devide_data(dataMat,labelMat)
print '训练数据：'+str(len(dataMat))
print '测试数据：'+str(len(testx))
xMat = mat(dataMat); yMat = mat(labelMat).T
xTx = xMat.T*xMat
if linalg.det(xTx) == 0.0:
print '没有解'
return
ws = xTx.I * (xMat.T*yMat)
return drawtest(dataMat,labelMat,ws.getA1(),page_name,hotnum,timeline,testx,testy,percent=percent,xlabelinput=['share_hourall','share_hotfan_per'],titleinput='shareall and sharehotfan:share_count',notshow = notshow)
