def extractSubEps(self, prefix):
'''
extract sub episodes of this episode recursively.
'''
sub_epsset = EpisodeSet([])
if prefix.length == 0:
for i in range(self.length):
pfx = [self.events[i]]
pep = SerialEpisode(pfx) # prefix episode
#print pep.repr
sub_epsset.add(pep)
#sub_eps[pep.repr] = sub_eps[pep.repr]+1 if sub_eps.has_key(pep.repr) else 1
sep = SerialEpisode(
self.events[i + 1:self.length]) # suffix episodes
sub_epsset.merge(sep.extractSubEps(pep))
else:
for i in range(self.length):
pfx = copy.deepcopy(prefix.events)
pfx.extend([self.events[i]])
pep = SerialEpisode(pfx) # prefix episode
#print pep.repr
sub_epsset.add(pep)
#sub_eps[pep.repr] = sub_eps[pep.repr]+1 if sub_eps.has_key(pep.repr) else 1
sep = SerialEpisode(
self.events[i + 1:self.length]) # suffix episodes
sub_epsset.merge(sep.extractSubEps(pep))
return sub_epsset
def extractSubEps(self,prefix):
'''
extract sub episodes of this episode recursively.
'''
sub_epsset = EpisodeSet([])
if prefix.length == 0:
for i in range(self.length):
pfx = [ self.events[i] ]
pep = SerialEpisode(pfx) # prefix episode
#print pep.repr
sub_epsset.add(pep)
#sub_eps[pep.repr] = sub_eps[pep.repr]+1 if sub_eps.has_key(pep.repr) else 1
sep = SerialEpisode(self.events[i+1:self.length]) # suffix episodes
sub_epsset.merge(sep.extractSubEps(pep))
else:
for i in range(self.length):
pfx = copy.deepcopy(prefix.events)
pfx.extend([ self.events[i] ])
pep = SerialEpisode(pfx) # prefix episode
#print pep.repr
sub_epsset.add(pep)
#sub_eps[pep.repr] = sub_eps[pep.repr]+1 if sub_eps.has_key(pep.repr) else 1
sep = SerialEpisode(self.events[i+1:self.length]) # suffix episodes
sub_epsset.merge(sep.extractSubEps(pep))
return sub_epsset
def __init__(self, interval, window):
"""
Constructor
"""
# Calling the super
super(Tsbatcher, self).__init__()
# Storing values
self.length = 0
self.interval = interval
self.starttime = 0
self.endtime = self.starttime
self.window = window
self.winIndex = -1
self.epsset = EpisodeSet([])
def run(self):
"""
Read the data in window, get episodes and their statistics information
into an episode set.
"""
global streamers,miners,conds,windows,win_epssets,nwinm
epsset = EpisodeSet()
while True:
if conds[self.name].acquire():
if self.name in windows.keys() and epsset.length == 0:
#start = time.time()
tsbatcher = Tsbatcher(self.mtp,windows[self.name])
epsset = tsbatcher.episodeStatistic()
#end = time.time()
#logging.critical('Time spent mining: %f seconds' % (end - start))
del windows[self.name]
print "\tMiner "+self.name+":"+"window is done."
conds[self.name].notify()
elif not streamers[self.name].is_alive():
conds[self.name].release()
break
conds[self.name].release()
else:
conds[self.name].wait()
if mutex.acquire():
if epsset.length != 0 and self.name not in win_epssets.keys():
win_epssets[self.name] = copy.deepcopy(epsset)
#win_epssets[self.name].__show__()
epsset = EpisodeSet()
nwinm[self.name] += 1
mutex.release()
time.sleep(1)
print "\tMiner "+self.name+":"+"is over."
def __init__(self, interval, window):
"""
Constructor
"""
# Calling the super
super(Tsbatcher, self).__init__()
# Storing values
self.length = 0
self.interval = interval
self.starttime = 0
self.endtime = self.starttime
self.window = window
self.winIndex = -1
self.epsset = EpisodeSet([])
e6 = Event("A,2")
e7 = Event("C,3")
e8 = Event("B,4")
e9 = Event("C,4")
e10 = Event("D,4")
el1 = [e1,e2,e3,e4]
el2 = [e3,e4,e5]
el3 = [e2,e6,e7]
el4 = [e6,e7,e8,e8]
print el1
print el2
print el3
print el4
eps1 = SerialEpisode(el1)
eps2 = SerialEpisode(el2)
eps3 = SerialEpisode(el3)
eps4 = SerialEpisode(el4)
epss1 = EpisodeSet([eps1,eps2])
epss2 = EpisodeSet([eps1,eps4])
epss1.__show__()
epss2.__show__()
print epss1.commonRatio(epss2)
class Tsbatcher(list):
def __init__(self, interval, window):
"""
Constructor
"""
# Calling the super
super(Tsbatcher, self).__init__()
# Storing values
self.length = 0
self.interval = interval
self.starttime = 0
self.endtime = self.starttime
self.window = window
self.winIndex = -1
self.epsset = EpisodeSet([])
######################################################################
# Generate Episodes
######################################################################
def episodeStatistic(self):
'''
Generate episode set in current Window with this time span batch.
'''
while len(self.window) + self.winIndex >= 0:
self.move()
seq = [ e for e in self ]
#print seq
serialepisode = SerialEpisode(seq)
self.epsset.merge(serialepisode.subeps())
return self.epsset
def extractEpisodes(self,prefix,sequence):
'''
Generate episodes in current time span batch.
'''
if len(prefix) == 0:
for i in range(len(sequence)):
pfx = [ sequence[i] ]
print pfx
tsq = copy.deepcopy(sequence[i+1:len(sequence)])
self.extractEpisodes(pfx,tsq)
else:
for i in range(len(sequence)):
pfx = copy.deepcopy(prefix)
pfx.extend([ sequence[i] ])
print pfx
tsq = copy.deepcopy(sequence[i+1:len(sequence)])
self.extractEpisodes(pfx,tsq)
######################################################################
# Management
######################################################################
def move(self):
"""
Move the batch along the window.
"""
out = None
if len(self) > 0:
out = self[-1]
del self[-1]
if len(self) == 0:
self.starttime += 1
else:
self.starttime = self[-1].time
#event = Event("X,0")
while len(self.window) + self.winIndex >= 0:
event = self.window[self.winIndex]
if len(self) == 0:
self.insert(0, event)
self.starttime = event.time
self.endtime = event.time
self.winIndex -= 1
else:
if event.time - self.starttime <= self.interval:
self.endtime = event.time
self.insert(0,event)
self.winIndex -= 1
else:
break
#print "batch:"+str(self)
def filled(self):
"""
Returns whether the window is full (i.e. contains length
elements).
"""
return len(self) == self.length
######################################################################
# Frequency
######################################################################
def getCount(self, event):
"""
Returns the count of the given event in this window.
"""
# Attempt 2: Still too slow
count = 0
for mEvent in self:
if event.__st__(mEvent):
count += 1
return count
# Attempt 1: Too slow
#return reduce((lambda x, y: x+y),
# map((lambda i: itemset <= i), self))
######################################################################
# Getters
######################################################################
def getStreamEventTypes(self):
return self.streamEventTypes
class Tsbatcher(list):
def __init__(self, interval, window):
"""
Constructor
"""
# Calling the super
super(Tsbatcher, self).__init__()
# Storing values
self.length = 0
self.interval = interval
self.starttime = 0
self.endtime = self.starttime
self.window = window
self.winIndex = -1
self.epsset = EpisodeSet([])
######################################################################
# Generate Episodes
######################################################################
def episodeStatistic(self):
'''
Generate episode set in current Window with this time span batch.
'''
while len(self.window) + self.winIndex >= 0:
self.move()
seq = [e for e in self]
#print seq
serialepisode = SerialEpisode(seq)
self.epsset.merge(serialepisode.subeps())
return self.epsset
def extractEpisodes(self, prefix, sequence):
'''
Generate episodes in current time span batch.
'''
if len(prefix) == 0:
for i in range(len(sequence)):
pfx = [sequence[i]]
print pfx
tsq = copy.deepcopy(sequence[i + 1:len(sequence)])
self.extractEpisodes(pfx, tsq)
else:
for i in range(len(sequence)):
pfx = copy.deepcopy(prefix)
pfx.extend([sequence[i]])
print pfx
tsq = copy.deepcopy(sequence[i + 1:len(sequence)])
self.extractEpisodes(pfx, tsq)
######################################################################
# Management
######################################################################
def move(self):
"""
Move the batch along the window.
"""
out = None
if len(self) > 0:
out = self[-1]
del self[-1]
if len(self) == 0:
self.starttime += 1
else:
self.starttime = self[-1].time
#event = Event("X,0")
while len(self.window) + self.winIndex >= 0:
event = self.window[self.winIndex]
if len(self) == 0:
self.insert(0, event)
self.starttime = event.time
self.endtime = event.time
self.winIndex -= 1
else:
if event.time - self.starttime <= self.interval:
self.endtime = event.time
self.insert(0, event)
self.winIndex -= 1
else:
break
#print "batch:"+str(self)
def filled(self):
"""
Returns whether the window is full (i.e. contains length
elements).
"""
return len(self) == self.length
######################################################################
# Frequency
######################################################################
def getCount(self, event):
"""
Returns the count of the given event in this window.
"""
# Attempt 2: Still too slow
count = 0
for mEvent in self:
if event.__st__(mEvent):
count += 1
return count
# Attempt 1: Too slow
#return reduce((lambda x, y: x+y),
# map((lambda i: itemset <= i), self))
######################################################################
# Getters
######################################################################
def getStreamEventTypes(self):
return self.streamEventTypes
e5 = Event("A,5")
e6 = Event("A,2")
e7 = Event("C,3")
e8 = Event("B,4")
e9 = Event("C,4")
e10 = Event("D,4")
el1 = [e1, e2, e3, e4]
el2 = [e3, e4, e5]
el3 = [e2, e6, e7]
el4 = [e6, e7, e8, e8]
print el1
print el2
print el3
print el4
eps1 = SerialEpisode(el1)
eps2 = SerialEpisode(el2)
eps3 = SerialEpisode(el3)
eps4 = SerialEpisode(el4)
epss1 = EpisodeSet([eps1, eps2])
epss2 = EpisodeSet([eps1, eps4])
epss1.__show__()
epss2.__show__()
print epss1.commonRatio(epss2)
