def plotDegreeDistribution(self,degreeList,GameCount,dataType, gameType):
p = Preprocess()
cachePath = '/home/yotoo/Project/comparison/'
plotPath = '/home/yotoo/Project/logPlot/'
# details about the data
maximum = degreeList[len(degreeList)-1]
dist = {}
for d in degreeList:
for i in range(1,maximum+1):
if (d == i):
if(dist.has_key(i)==False):
dist[i] = 1
else:
degree = dist[i]
degree = degree + 1
dist[i] = degree
else:
continue
print dist.items()
# plot degree distribution
degrees = dist.keys()
players = dist.values()
#details about this distribution
sortedP = dist.values()
sortedP.sort()
sortedD = dist.keys()
sortedD.sort()
minimum = sortedP[0]
sortedP.reverse()
maximum = sortedP[0]
playerCount = len(players)
degreeCount = len(degrees)
medianIndex = (minimum + degreeCount) / 2
Q1Index = (minimum + medianIndex) / 2
Q3Index = (medianIndex + degreeCount) / 2
median = sortedD[medianIndex]
Q1 = sortedD[Q1Index]
Q3 = sortedD[Q3Index]
kurtosis = self.getKurtosisSkewness(degrees)
plotText = 'Total Games: %d \nTotal Players: %d \nMaximum Degree: %d \nQ3 Degree: %d\nMedian Degree: %d \nQ1 Degree: %d \nMinimum Degree: %d \nKurtosis: %.2e' %(GameCount,playerCount,maximum,Q3,median,Q1,minimum,kurtosis)
#dataType
if (dataType == 'Real Data'):
savePath = plotPath + 'real_data/'
if (gameType == '19+ranked'):
title = 'Degree Distribution(Border size: 19; Game Type: Ranked)'
# # original
plt.grid(True)
plt.xlabel('Degree')
plt.ylabel('Players')
plt.title(title)
plt.plot(degrees,players,'ro',label = plotText)
plt.legend()
plt.savefig(savePath + title+' original' + ".png")
plt.show()
# # log degree
plt.grid(True)
plt.xlabel('Log Degree')
plt.ylabel('Players')
plt.title(title)
plt.semilogx(degrees,players,'ro',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log x' + ".png")
plt.show()
# # log players
plt.grid(True)
plt.xlabel('Degree')
plt.ylabel(' Log Players')
plt.title(title)
plt.semilogy(degrees,players,'ro',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log y' + ".png")
plt.show()
# # log degrees and players
plt.grid(True)
plt.xlabel('Log Degree')
plt.ylabel('Log Players')
plt.title(title)
plt.loglog(degrees,players,'ro',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log x y' + ".png")
plt.show()
# # plot 4-up images for original, log x and y, log x, log y
# original
plt.subplot(221)
plt.plot(degrees,players,'ro')
plt.title('Original Plot')
plt.grid(True)
#log degrees
plt.subplot(222)
plt.semilogx(degrees,players,'ro')
plt.title('Log Degrees')
plt.grid(True)
#log players
plt.subplot(223)
plt.semilogy(degrees,players,'ro')
plt.title('Log Players')
plt.grid(True)
#log degrees and players
plt.subplot(224)
plt.loglog(degrees,players,'ro')
plt.title('Log Degrees and Players')
plt.grid(True)
plt.savefig(savePath + "random_data_19_ranked_4_ups.png")
plt.show()
p.save(dist,cachePath+title+'.txt')
elif (gameType == '19+free'):
title = 'Degree Distribution(Border size: 19; Game Type: Free)'
# # original
plt.grid(True)
plt.xlabel('Degree')
plt.ylabel('Players')
plt.title(title)
plt.plot(degrees,players,'gs',label = plotText)
plt.legend()
plt.savefig(savePath + title+' original' + ".png")
plt.show()
# # log degree
plt.grid(True)
plt.xlabel('Log Degree')
plt.ylabel('Players')
plt.title(title)
plt.semilogx(degrees,players,'gs',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log x' + ".png")
plt.show()
# # log players
plt.grid(True)
plt.xlabel('Degree')
plt.ylabel(' Log Players')
plt.title(title)
plt.semilogy(degrees,players,'gs',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log y' + ".png")
plt.show()
# # log degrees and players
plt.grid(True)
plt.xlabel('Log Degree')
plt.ylabel('Log Players')
plt.title(title)
plt.loglog(degrees,players,'gs',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log x y' + ".png")
plt.show()
# # plot 4-up images for original, log x and y, log x, log y
# original
plt.subplot(221)
plt.plot(degrees,players,'gs')
plt.title('Original Plot')
plt.grid(True)
#log degrees
plt.subplot(222)
plt.semilogx(degrees,players,'gs')
plt.title('Log Degrees')
plt.grid(True)
#log players
plt.subplot(223)
plt.semilogy(degrees,players,'sg')
plt.title('Log Players')
plt.grid(True)
#log degrees and players
plt.subplot(224)
plt.loglog(degrees,players,'sg')
plt.title('Log Degrees and Players')
plt.grid(True)
plt.savefig(savePath + "random_data_19_free_4_ups.png")
plt.show()
p.save(dist,cachePath+title+'.txt')
elif (gameType == '9+free'):
title = 'Degree Distribution(Border size: 9; Game Type: Free)'
# # original
plt.grid(True)
plt.xlabel('Degree')
plt.ylabel('Players')
plt.title(title)
plt.plot(degrees,players,'b^',label = plotText)
plt.legend()
plt.savefig(savePath + title+' original' + ".png")
plt.show()
# # log degree
plt.grid(True)
plt.xlabel('Log Degree')
plt.ylabel('Players')
plt.title(title)
plt.semilogx(degrees,players,'b^',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log x' + ".png")
plt.show()
# # log players
plt.grid(True)
plt.xlabel('Degree')
plt.ylabel(' Log Players')
plt.title(title)
plt.semilogy(degrees,players,'b^',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log y' + ".png")
plt.show()
# # log degrees and players
plt.grid(True)
plt.xlabel('Log Degree')
plt.ylabel('Log Players')
plt.title(title)
plt.loglog(degrees,players,'b^',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log x y' + ".png")
plt.show()
# # plot 4-up images for original, log x and y, log x, log y
# original
plt.subplot(221)
plt.plot(degrees,players,'b^')
plt.title('Original Plot')
plt.grid(True)
#log degrees
plt.subplot(222)
plt.semilogx(degrees,players,'b^')
plt.title('Log Degrees')
plt.grid(True)
#log players
plt.subplot(223)
plt.semilogy(degrees,players,'b^')
plt.title('Log Players')
plt.grid(True)
#log degrees and players
plt.subplot(224)
plt.loglog(degrees,players,'b^')
plt.title('Log Degrees and Players')
plt.grid(True)
plt.savefig(savePath + "random_data_9_free_4_ups.png")
plt.show()
p.save(dist,cachePath+title+'.txt')
elif (dataType == 'Random Data'):
savePath = plotPath + '/random_data/'
if (gameType == '19+ranked'):
title = 'Degree Distribution(Border size: 19; Game Type: Ranked) *random '
# # original
plt.grid(True)
plt.xlabel('Degree')
plt.ylabel('Players')
plt.title(title)
plt.plot(degrees,players,'ro',label = plotText)
plt.legend()
plt.savefig(savePath + title+' original' + ".png")
plt.show()
# # log degree
plt.grid(True)
plt.xlabel('Log Degree')
plt.ylabel('Players')
plt.title(title)
plt.semilogx(degrees,players,'ro',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log x' + ".png")
plt.show()
# # log players
plt.grid(True)
plt.xlabel('Degree')
plt.ylabel(' Log Players')
plt.title(title)
plt.semilogy(degrees,players,'ro',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log y' + ".png")
plt.show()
# # log degrees and players
plt.grid(True)
plt.xlabel('Log Degree')
plt.ylabel('Log Players')
plt.title(title)
plt.loglog(degrees,players,'ro',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log x y' + ".png")
plt.show()
# # plot 4-up images for original, log x and y, log x, log y
# original
plt.subplot(221)
plt.plot(degrees,players,'ro')
plt.title('Original Plot')
plt.grid(True)
#log degrees
plt.subplot(222)
plt.semilogx(degrees,players,'ro')
plt.title('Log Degrees')
plt.grid(True)
#log players
plt.subplot(223)
plt.semilogy(degrees,players,'ro')
plt.title('Log Players')
plt.grid(True)
#log degrees and players
plt.subplot(224)
plt.loglog(degrees,players,'ro')
plt.title('Log Degrees and Players')
plt.grid(True)
plt.savefig(savePath + "random_data_19_ranked_4_ups.png")
plt.show()
p.save(dist,cachePath+title+'.txt')
elif (gameType == '19+free'):
title = 'Degree Distribution(Border size: 19; Game Type: Free) *random '
# # original
plt.grid(True)
plt.xlabel('Degree')
plt.ylabel('Players')
plt.title(title)
plt.plot(degrees,players,'gs',label = plotText)
plt.legend()
plt.savefig(savePath + title+' original' + ".png")
plt.show()
# # log degree
plt.grid(True)
plt.xlabel('Log Degree')
plt.ylabel('Players')
plt.title(title)
plt.semilogx(degrees,players,'gs',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log x' + ".png")
plt.show()
# # log players
plt.grid(True)
plt.xlabel('Degree')
plt.ylabel(' Log Players')
plt.title(title)
plt.semilogy(degrees,players,'gs',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log y' + ".png")
plt.show()
# # log degrees and players
plt.grid(True)
plt.xlabel('Log Degree')
plt.ylabel('Log Players')
plt.title(title)
plt.loglog(degrees,players,'gs',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log x y' + ".png")
plt.show()
# # plot 4-up images for original, log x and y, log x, log y
# original
plt.subplot(221)
plt.plot(degrees,players,'gs')
plt.title('Original Plot')
plt.grid(True)
#log degrees
plt.subplot(222)
plt.semilogx(degrees,players,'gs')
plt.title('Log Degrees')
plt.grid(True)
#log players
plt.subplot(223)
plt.semilogy(degrees,players,'sg')
plt.title('Log Players')
plt.grid(True)
#log degrees and players
plt.subplot(224)
plt.loglog(degrees,players,'sg')
plt.title('Log Degrees and Players')
plt.grid(True)
plt.savefig(savePath + "random_data_19_free_4_ups.png")
plt.show()
p.save(dist,cachePath+title+'.txt')
elif (gameType == '9+free'):
title = 'Degree Distribution(Border size: 9; Game Type: Free) *random '
# # original
plt.grid(True)
plt.xlabel('Degree')
plt.ylabel('Players')
plt.title(title)
plt.plot(degrees,players,'b^',label = plotText)
plt.legend()
plt.savefig(savePath + title+' original' + ".png")
plt.show()
# # log degree
plt.grid(True)
plt.xlabel('Log Degree')
plt.ylabel('Players')
plt.title(title)
plt.semilogx(degrees,players,'b^',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log x' + ".png")
plt.show()
# # log players
plt.grid(True)
plt.xlabel('Degree')
plt.ylabel(' Log Players')
plt.title(title)
plt.semilogy(degrees,players,'b^',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log y' + ".png")
plt.show()
# # log degrees and players
plt.grid(True)
plt.xlabel('Log Degree')
plt.ylabel('Log Players')
plt.title(title)
plt.loglog(degrees,players,'b^',label = plotText)
plt.legend()
plt.savefig(savePath + title+' log x y' + ".png")
plt.show()
# # plot 4-up images for original, log x and y, log x, log y
# original
plt.subplot(221)
plt.plot(degrees,players,'b^')
plt.title('Original Plot')
plt.grid(True)
#log degrees
plt.subplot(222)
plt.semilogx(degrees,players,'b^')
plt.title('Log Degrees')
plt.grid(True)
#log players
plt.subplot(223)
plt.semilogy(degrees,players,'b^')
plt.title('Log Players')
plt.grid(True)
#log degrees and players
plt.subplot(224)
plt.loglog(degrees,players,'b^')
plt.title('Log Degrees and Players')
plt.grid(True)
plt.savefig(savePath + "random_data_9_free_4_ups.png")
plt.show()
p.save(dist,cachePath+title+'.txt')
def main():
#############################################################
# Real Code
# # initialization
startTime = time.time()
process = Preprocess()
relation = Relationship()
comparison = Comparison()
clustering = Clustering()
#--------------------------------Preprocess Data-----------------------------------------------------------
# data = P.load('cleandata.txt')
# cleandata = P.preprocess(data,'19+ranked')
# randomData = P.createRandomData(cleandata,50000)
# P.save(cleandata,'prepro_data_19_ranked.txt')
# cleandata = P.preprocess(data,'19+free')
# P.save(cleandata,'prepro_data_19_free.txt')
# cleandata = P.preprocess(data,'9+ranked')
# P.save(cleandata,'prepro_data_9_ranked.txt')
# cleandata = P.preprocess(data,'9+free')
# P.save(cleandata,'prepro_data_9_free.txt')
# # create random data
# cleandata = P.load('prepro_data_19_ranked.txt')
# randomData = P.getSubset(cleandata,50000,'Random')
# P.save(randomData,'prepro_random_data_19_ranked.txt')
# cleandata = P.load('prepro_data_19_free.txt')
# randomData = P.getSubset(cleandata,50000,'Random')
# P.save(randomData,'prepro_random_data_19_free.txt')
# cleandata = P.load('prepro_data_9_free.txt')
# randomData = P.getSubset(cleandata,50000,'Random')
# P.save(randomData,'prepro_random_data_9_free.txt')
#-----------------------------------Degree Distribution-----------------------------------------------------------
# data = process.load('prepro_data_19_ranked.txt')
# relationship = relation.findRelationship(data)
# gameCount = len(data)
# degreeList = relation.cleanRelationship(relationship,'DegreeDistribution')
# relation.plotDegreeDistribution(degreeList,gameCount, 'Real Data', '19+ranked')
# data = process.load('prepro_data_19_free.txt')
# relationship = relation.findRelationship(data)
# gameCount = len(data)
# degreeList = relation.cleanRelationship(relationship,'DegreeDistribution')
# relation.plotDegreeDistribution(degreeList,gameCount,'Real Data','19+free')
# # can not find records match "border size: 9 & game type: free"
# data = process.load('prepro_data_9_free.txt')
# relationship = relation.findRelationship(data)
# gameCount = len(data)
# degreeList = relation.cleanRelationship(relationship,'DegreeDistribution')
# relation.plotDegreeDistribution(degreeList,gameCount,'Real Data','9+free')
# # # random data
# data = process.load('prepro_random_data_19_ranked.txt')
# relationship = relation.findRelationship(data)
# gameCount = len(data)
# degreeList = relation.cleanRelationship(relationship,'DegreeDistribution')
# relation.plotDegreeDistribution(degreeList,gameCount,'Random Data','19+ranked')
# data = process.load('prepro_random_data_19_free.txt')
# relationship = relation.findRelationship(data)
# gameCount = len(data)
# degreeList = relation.cleanRelationship(relationship,'DegreeDistribution')
# relation.plotDegreeDistribution(degreeList,gameCount,'Random Data','19+free')
# # # can not find records match "border size: 9 & game type: free"
# data = process.load('prepro_random_data_9_free.txt')
# relationship = relation.findRelationship(data)
# gameCount = len(data)
# degreeList = relation.cleanRelationship(relationship,'DegreeDistribution')
# relation.plotDegreeDistribution(degreeList,gameCount,'Random Data','9+free')
#------------------------------------Shortest Path-------------------------------------------------------------------
# data = P.load('prepro_data_19_ranked.txt')
# relationship = R.findRelationship(data)
# playerList = relationship.keys()
# playerList.sort()
# P.save(playerList,'playerList_19_ranked.txt')
# pathMatrix = R.createPathMatrix(relationship,playerList)
# print pathMatrix
# data = P.load('prepro_data_19_free.txt')
# relationship = R.findRelationship(data)
# playerList = relationship.keys()
# playerList.sort()
# P.save(playerList,'playerList_19_free.txt')
# pathMatrix = R.createPathMatrix(relationship,playerList)
# print pathMatrix
# data = P.load('prepro_data_9_free.txt')
# relationship = R.findRelationship(data)
# playerList = relationship.keys()
# playerList.sort()
# P.save(playerList,'playerList_9_free.txt')
# pathMatrix = R.createPathMatrix(relationship,playerList)
# print pathMatrix
# # random data
# data = P.load('prepro_random_data_19_ranked.txt')
# relationship = R.findRelationship(data)
# playerList = relationship.keys()
# playerList.sort()
# P.save(playerList,'playerList__random_19_ranked.txt')
# pathMatrix = R.createPathMatrix(relationship,playerList)
# print pathMatrix
# P.save(pathMatrix,'pathMatrix_19_ranked.txt')
# data = P.load('prepro_random_data_19_free.txt')
# relationship = R.findRelationship(data)
# playerList = relationship.keys()
# playerList.sort()
# P.save(playerList,'playerList_random_19_free.txt')
# pathMatrix = R.createPathMatrix(relationship,playerList)
# print pathMatrix
# P.save(pathMatrix,'pathMatrix_19_free.txt')
# data = P.load('prepro_random_data_9_free.txt')
# relationship = R.findRelationship(data)
# playerList = relationship.keys()
# playerList.sort()
# P.save(playerList,'playerList_random_9_free.txt')
# pathMatrix = R.createPathMatrix(relationship,playerList)
# print pathMatrix
# P.save(pathMatrix,'pathMatrix_9_free.txt')
#------------------------------------Clustering-------------------------------------------------------------------
# data = P.load('prepro_random_data_19_ranked.txt')
# relationship = R.findRelationship(data)
# gameCount = len(data)
# playerList = relationship.keys()
# C = Clustering()
# clusterDict = C.cluster(playerList,relationship)
# P.save(clusterDict,'Clustering Result_random_data_19_ranked.txt')
# data = P.load('prepro_random_data_19_free.txt')
# relationship = R.findRelationship(data)
# gameCount = len(data)
# playerList = relationship.keys()
# C = Clustering()
# clusterDict = C.cluster(playerList,relationship)
# P.save(clusterDict,'Clustering Result_random_data_19_free.txt')
# data = P.load('prepro_random_data_9_free.txt')
# relationship = R.findRelationship(data)
# gameCount = len(data)
# playerList = relationship.keys()
# C = Clustering()
# clusterDict = C.cluster(playerList,relationship)
# P.save(clusterDict,'Clustering Result_random_data_9_free.txt')
# data = P.load('test_data.txt')
# relationship = R.findRelationship(data)
# gameCount = len(data)
# playerList = relationship.keys()
# C = Clustering()
# clusterDict = C.cluster(playerList,relationship)
# print clusterDict
# print type(clusterDict)
# C.getIntelView(clusterDict)
#------------------------------------Comparison-------------------------------------------------------------------
# P.save(clusterDict,'Clustering Result_test_data.txt')
# data = process.load('/home/yotoo/Project/comparison/Degree Distribution(Border size: 9; Game Type: Free).txt')
data = process.load('/home/yotoo/Project/comparison/Degree Distribution(Border size: 19; Game Type: Ranked).txt')
flag = '19 + Ranked'
comparison.linearRegression(data,flag)
data = process.load('/home/yotoo/Project/comparison/Degree Distribution(Border size: 19; Game Type: Free).txt')
flag = '19 + Free'
comparison.linearRegression(data,flag)
data = process.load('/home/yotoo/Project/comparison/Degree Distribution(Border size: 9; Game Type: Free).txt')
flag = '9 + Free'
comparison.linearRegression(data,flag)
endTime = time.time()
print 'totally use ' + str(endTime-startTime) + ' seconds!'