def allshare(page_name,hours,justgetdata=None):
import datetime
import numpy as np
from draw_corrcoef import draw_corre
fbsql = fb_mysql()
if page_name == 'all':
alllist = []
for i in pagelist:
alllist += fbsql.defind_by_self('select user_msg.msgid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromid = \'%s\')'%i)
else:
alllist = fbsql.defind_by_self('select user_msg.msgid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromname=\''+page_name+'\'')
dicttoid = {}
msgdict = {}
onehoursdict = {}
for i in alllist:
if i[0] not in msgdict:
msgdict[i[0]] = fbsql.defind_by_self('select share_count,created_time from user_msg where msgid=\'%s\''%i[0])[0]
temptime = msgdict[i[0]][1]+datetime.timedelta(hours=hours)
strtime = temptime.strftime("%Y-%m-%d %H:%M:%S")
onehoursdict[i[0]] = fbsql.defind_by_self('select toid from share_fb_msg_new where trend_mid=\'%s\' and created_time<\'%s\''%(i[0],strtime))
if not onehoursdict[i[0]]:
temptime = msgdict[i[0]][1]+datetime.timedelta(hours=hours+3)
strtime = temptime.strftime("%Y-%m-%d %H:%M:%S")
onehoursdict[i[0]] = fbsql.defind_by_self('select toid from share_fb_msg_new where trend_mid=\'%s\' and created_time<\'%s\''%(i[0],strtime))
dictx = []
dicty = []
allcount = 0
for i in msgdict:
if onehoursdict[i] and msgdict[i][0]:
allcount += 1
dictx.append(len(onehoursdict[i]))
dicty.append(msgdict[i][0])
if justgetdata:
return (onehoursdict,msgdict)
testdatax = []
testdatay = []
dicttrainx = []
dicttrainy = []
argrmdict = dict()
arragerdm = np.random.choice(len(dictx),int(len(dictx)*0.25),replace=False)
for i in arragerdm:
argrmdict[i] = None
for i in range(len(dictx)):
if i in argrmdict:
testdatax.append(dictx[i])
testdatay.append(dicty[i])
else:
dicttrainx.append(dictx[i])
dicttrainy.append(dicty[i])
print '总数据：'+str(allcount)
print '训练数据：'+str(len(dicttrainx))
print '测试数据：'+str(len(testdatax))
xlabel = str(hours)+'_hours get_count'
ylabel = 'share_count'
title = page_name+' '+str(hours)+'_hour:all_time_true_share_count'
return draw_corre(xdata=np.log10(dicttrainx),ydata=np.log10(dicttrainy),xlabelstr=xlabel, ylabelstr=ylabel, titlestr=title, testdatax=np.log10(testdatax),testdatay = np.log10(testdatay))
def drawhotfun(hotfan,onehoursdict,msgdict,page_name,hotfannum,hours,percent,hotnot):
import numpy as np
from draw_corrcoef import draw_corre
dictx = []
dicty = []
for i in onehoursdict.keys():
tempx = 0
if onehoursdict[i]:
if len(onehoursdict[i])<len(hotfan):
for j in onehoursdict[i]:
if j[0] in hotfan:
tempx += 1
else:
for j in hotfan:
if (j,) in onehoursdict[i]:
tempx += 1
if not hotnot:
x = tempx/len(onehoursdict[i])
else:
x = (len(onehoursdict[i])-tempx)/len(onehoursdict[i])
if msgdict[i][0] and x > 0:
y = msgdict[i][0]
dictx.append(x)
dicty.append(y)
testdatax = []
testdatay = []
dicttrainx = []
dicttrainy = []
argrmdict = dict()
arragerdm = np.random.choice(len(dictx),int(len(dictx)*0.25),replace=False)
for i in arragerdm:
argrmdict[i] = None
for i in range(len(dictx)):
if i in argrmdict:
testdatax.append(dictx[i])
testdatay.append(dicty[i])
else:
dicttrainx.append(dictx[i])
dicttrainy.append(dicty[i])
print '训练数据：'+str(len(dicttrainx))
print '测试数据：'+str(len(testdatax))
xlabel = 'hot_fan_p'
ylabel = 'share_count'
title = page_name+' fan_difind:share_'+str(hotfannum)+percent+'up time:'+str(hours-8)+'h'
return draw_corre(xdata=np.log10(dicttrainx), ydata=np.log10(dicttrainy), xlabelstr=xlabel, ylabelstr=ylabel, titlestr=title, testdatax=np.log10(testdatax), testdatay=np.log10(testdatay))
def drawtest(dataMat,labelMat,ws,page_name,hotnum,timeline,testx,testy,percent='',xlabelinput='',titleinput='',notshow = None):
import numpy as np
from draw_corrcoef import draw_corre
testdatax = []
testdatay = []
dictx = []
dicty = []
for i in range(len(dataMat)):
tempx = dataMat[i]
x = ws[0]*tempx[0] + ws[1]*tempx[1] + ws[2]
y = labelMat[i]
dictx.append(x)
dicty.append(y)
for i in range(len(testx)):
tempx = testx[i]
x = ws[0]*tempx[0] + ws[1]*tempx[1] + ws[2]
y = testy[i]
testdatax.append(x)
testdatay.append(y)
xlabel = str(ws[0])+'*'+xlabelinput[0]+'+'+str(ws[1])+'*'+xlabelinput[1]+'+'+str(ws[2])+' '+str(timeline-8)+'h'
ylabel = 'share_count'
title = page_name+' '+titleinput+' time:'+str(timeline-8)+'h fan_defind:'+str(hotnum)+percent+'up'
return draw_corre(xdata=dictx,ydata=dicty,xlabelstr=xlabel,ylabelstr=ylabel,titlestr=title,testdatax=testdatax,testdatay = testdatay,notshow=notshow)
def likesum_nothotfan(page_name,hotnum,timeline,percent = '',justgetdata=None):
import numpy as np
import datetime
from draw_corrcoef import draw_corre
fbsql = fb_mysql()
if page_name == 'all':
alllist = []
hotfan = []
for i in pagelist:
eachlist = fbsql.defind_by_self('select user_msg.msgid,share_fb_msg_new.fromid,toid,share_fb_msg_new.created_time,share_fb_msg_new.like_count from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromid = \'%s\')'%i)
alllist += eachlist
dicttoid = {}
for j in eachlist:
if j[2] in dicttoid:
dicttoid[j[2]] += 1
else:
dicttoid[j[2]] = 1
hotfan = dict(hotfan,**percentfan(hotnum,dicttoid,percent))
else:
alllist = fbsql.defind_by_self('select user_msg.msgid,share_fb_msg_new.fromid,toid,share_fb_msg_new.created_time,share_fb_msg_new.like_count from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromname=\''+page_name+'\'')
dicttoid = {}
for i in alllist:
if i[2] in dicttoid:
dicttoid[i[2]] += 1
else:
dicttoid[i[2]] = 1
hotfan = percentfan(hotnum,dicttoid,percent)
timedict = dict()
for i in alllist:
if i[0] not in timedict:
tempdict = {}
tempdetail = fbsql.defind_by_self('select share_count,created_time from user_msg where msgid=\'%s\''%i[0])[0]
tempdict['share_count'] = tempdetail[0]
tempdict['created_time'] = tempdetail[1]
tempdict['line_time']=tempdict['created_time']+datetime.timedelta(hours=timeline)
tempdict['all_num'] = 0
tempdict['like_sum'] = 0
tempdict['all_like_sum'] = 0
if not tempdict['share_count']:
continue
if i[3] < tempdict['line_time'] and i[1]!=i[2]:
tempdict['all_num'] = 1
if i[4] >0:
if i[2] not in hotfan:
tempdict['like_sum']+=i[4]
tempdict['all_like_sum']+=i[4]
timedict[i[0]] = tempdict
else:
tempdict = timedict[i[0]]
if i[3] < tempdict['line_time'] and i[1]!=i[2]:
tempdict['all_num'] += 1
if i[4] >0:
if i[2] not in hotfan:
tempdict['like_sum']+=i[4]
tempdict['all_like_sum']+=i[4]
dictx = []
dicty = []
if justgetdata:
return timedict
for i in timedict.keys():
print i+':'+str(timedict[i]['like_sum'])
if timedict[i]['all_num'] > 0 and timedict[i]['like_sum']>0:
x = timedict[i]['like_sum']
dictx.append(x)
dicty.append(timedict[i]['share_count'])
testdatax = []
testdatay = []
dicttrainx = []
dicttrainy = []
argrmdict = dict()
arragerdm = np.random.choice(len(dictx),int(len(dictx)*0.25),replace=False)
for i in arragerdm:
argrmdict[i] = None
for i in range(len(dictx)):
if i in argrmdict:
testdatax.append(dictx[i])
testdatay.append(dicty[i])
else:
dicttrainx.append(dictx[i])
dicttrainy.append(dicty[i])
print '训练数据：'+str(len(dicttrainx))
print '测试数据：'+str(len(testdatax))
xlabel='like_sum_notfan/all_num '+str(timeline)+'h'
ylabel='share_count'
title=page_name+' like_sum_notfan:share_count time:'+str(timeline-8)+'h fan_defind:'+str(hotnum)+percent+'up'
return draw_corre(xdata=np.log10(dicttrainx), ydata=np.log10(dicttrainy), xlabelstr=xlabel, ylabelstr=ylabel, titlestr=title, testdatax=np.log10(testdatax), testdatay=np.log10(testdatay))
