def KMeans2(TabData, ClusterCount, Metric, Weight, PrimCent):
TabDataNormal, UnKeysMat, MEMat, DMat, Y=PrepareForKNN(TabData, len(TabData[0]))
TabDataNormal.pop(0)
TabDataNormal=list(zip(*TabDataNormal))
mu1=PrimaryCenter(TabDataNormal, Y, ClusterCount, Metric, Weight, PrimCent)
Default=[]
for i in TabDataNormal[0]:
if isinstance(i, list)==True:
Default.append([0 for ii in i])
else:
Default.append(0)
while True:
ClusterList=CreateClusterList(TabDataNormal, mu1, Metric, Weight)
mu2=[]
#print('0: {0} 1: {1} 2: {2} 3: {3} 4: {4}'. format(len(ClusterList[0]), len(ClusterList[1]), len(ClusterList[2]), len(ClusterList[3]), len(ClusterList[4])))
for Cluster in ClusterList:
center=CenterOfMass(Cluster, Default)
mu2.append(center)
if (mu1==mu2):
break
else:
mu1=mu2[:]
mu=CloseToCenterOfMass(ClusterList, mu1, Metric, Weight)
OptTabData=GetOptimizedTabData(TabData, TabDataNormal, mu)
return OptTabData
def TimurAlgorithm2(TabData):
OptTabData, Id=[], {}
TabDataNormal, UnKeysMat, MEMat, DMat, Y=PrepareForKNN(TabData, len(TabData[0]))
TabDataNormal.pop(0)
ClusterList={}
for i, y in enumerate(Y):
if ClusterList.get(y)==None:
ClusterList[y]=[i]
else:
ClusterList[y].append(i)
for key in ClusterList.keys():
Cluster=ClusterList[key]
for Column in TabDataNormal:
min, max=Column[Cluster[0]], Column[Cluster[0]]
idmin, idmax=Cluster[0], Cluster[0]
for i in Cluster[1:]:
if Column[i]<min:
min=Column[i]
idmin=i
elif Column[i]>max:
max=Column[i]
idmax=i
Id[idmin], Id[idmax]=idmin, idmax
for id in sorted(Id.keys()):
OptTabData.append(TabData[id])
return OptTabData
def TimurAlgorithm(TabData):
IdTab=[Str[0] for Str in TabData]
TabDataNormal, UnKeysMat, MEMat, DMat, Y=PrepareForKNN(TabData, len(TabData[0]))
TabDataNormal.pop(0)
#TabDataNormal=list(zip(*TabDataNormal))
Id, d=[], {}
for y in Y:
if d.get(y)==None:
d[y]=len(d)
ClusterList=[[] for i in range(0, len(d))]
for i, y in enumerate(Y):
ClusterList[d.get(y)].append(i)
for Column in TabDataNormal:
for ClNum, Cluster in enumerate(ClusterList):
#flash('------'+str(ClNum)+'------')
max, min, idmax, idmin=Column[Cluster[0]], Column[Cluster[0]], Cluster[0], Cluster[0]
for i in Cluster:
if Column[i]>max:
max=Column[i]
idmax=i
if Column[i]<min:
min=Column[i]
idmin=i
#flash('IdMax: '+str(idmax)+' Max: '+str(max))
#flash('IdMin: '+str(idmin)+' Min: '+str(max))
Id.append(IdTab[idmax])
Id.append(IdTab[idmin])
Id.sort()
Id1 = [e for i,e in enumerate(Id) if e not in Id[:i]]
return Id1
def Classification2(TabData, Metric, Weight, Similarity):
TabDataNormal, UnKeysMat, MEMat, DMat, Y=PrepareForKNN(TabData, len(TabData[0]))
TabDataNormal.pop(0)
TabDataNormal=list(zip(*TabDataNormal))
for i, Str1 in enumerate(TabDataNormal):
PartTabDataNormal=(TabDataNormal)[i+1:]
for j, Str2 in enumerate(PartTabDataNormal):
dist=globals()[Metric](Str1, Str2, Weight)
if dist<=Similarity:
PartTabDataNormal.pop(j)
TabDataNormal.pop(i+1+j)
TabData.pop(i+1+j)
return TabData