def __init__(self):
""" Initializes the eLCS algorithm """
print("eLCS: Initializing Algorithm...")
#Global Parameters-------------------------------------------------------------------------------------
self.population = None # The rule population (the 'solution/model' evolved by eLCS)
self.learnTrackOut = None # Output file that will store tracking information during learning
#-------------------------------------------------------
# POPULATION REBOOT - Begin eLCS learning from an existing saved rule population
#-------------------------------------------------------
if cons.doPopulationReboot:
self.populationReboot()
#-------------------------------------------------------
# NORMAL eLCS - Run eLCS from scratch on given data
#-------------------------------------------------------
else:
try:
self.learnTrackOut = open(cons.outFileName+'_LearnTrack.txt','w')
except Exception as inst:
print(type(inst))
print(inst.args)
print(inst)
print('cannot open', cons.outFileName+'_LearnTrack.txt')
raise
else:
self.learnTrackOut.write("Explore_Iteration\tMacroPopSize\tMicroPopSize\tAccuracy_Estimate\tAveGenerality\tExpRules\tTime(min)\n")
# Instantiate Population---------
self.population = ClassifierSet()
self.exploreIter = 0
self.correct = [0.0 for i in range(cons.trackingFrequency)]
#Run the eLCS algorithm-------------------------------------------------------------------------------
self.results = self.run_eLCS()
def populationReboot(self):
""" Manages the reformation of a previously saved eLCS classifier population. """
#--------------------------------------------------------------------
try: #Re-open track learning file for continued tracking of progress.
self.learnTrackOut = open(cons.outFileName + '_LearnTrack.txt',
'a')
except Exception as inst:
print(type(inst))
print(inst.args)
print(inst)
print('cannot open', cons.outFileName + '_LearnTrack.txt')
raise
#Extract last iteration from file name---------------------------------------------
temp = cons.popRebootPath.split('_')
iterRef = len(temp) - 1
completedIterations = int(temp[iterRef])
print("Rebooting rule population after " + str(completedIterations) +
" iterations.")
self.exploreIter = completedIterations - 1
for i in range(len(cons.learningCheckpoints)):
cons.learningCheckpoints[i] += completedIterations
cons.maxLearningIterations += completedIterations
#Rebuild existing population from text file.--------
self.population = ClassifierSet(cons.popRebootPath)
def populationReboot(self):
""" Manages the reformation of a previously saved eLCS classifier population. """
#--------------------------------------------------------------------
try: #Re-open track learning file for continued tracking of progress.
self.learnTrackOut = open(cons.outFileName + '_LearnTrack.txt',
'a')
except Exception as inst:
print(type(inst))
print(inst.args)
print(inst)
print('cannot open', cons.outFileName + '_LearnTrack.txt')
raise
#Extract last iteration from file name---------------------------------------------
temp = cons.popRebootPath.split('_')
iterRef = len(temp) - 1
completedIterations = int(temp[iterRef])
print("Rebooting rule population after " + str(completedIterations) +
" iterations.")
self.exploreIter = completedIterations - 1
for i in range(len(cons.learningCheckpoints)):
cons.learningCheckpoints[i] += completedIterations
cons.maxLearningIterations += completedIterations
#Rebuild existing population from text file.--------
self.population = ClassifierSet(cons.popRebootPath)
#---------------------------------------------------
try: #Obtain correct track
f = open(cons.popRebootPath + "_PopStats.txt", 'r')
except Exception as inst:
print(type(inst))
print(inst.args)
print(inst)
print('cannot open', cons.outFileName + '_LearnTrack.txt')
raise
else:
correctRef = 26 #File reference position - only works with eLCS demo version 5 or 6 PopStats output files.
tempLine = None
for i in range(correctRef):
tempLine = f.readline()
tempList = tempLine.strip().split('\t')
self.correct = tempList
if cons.env.formatData.discretePhenotype:
for i in range(len(self.correct)):
self.correct[i] = int(self.correct[i])
else:
for i in range(len(self.correct)):
self.correct[i] = float(self.correct[i])
f.close()
def __init__(self):
""" Initializes the eLCS algorithm """
#Global Parameters-------------------------------------------------------------------------------------
self.population = None # The rule population (the 'solution/model' evolved by eLCS)
self.learnTrackOut = None # Output file that will store tracking information during learning
#-------------------------------------------------------
# POPULATION REBOOT - Begin eLCS learning from an existing saved rule population
#-------------------------------------------------------
if cons.doPopulationReboot:
self.populationReboot()
#-------------------------------------------------------
# NORMAL eLCS - Run eLCS from scratch on given data
#-------------------------------------------------------
else:
# Instantiate Population---------
self.population = ClassifierSet()
self.exploreIter = 0
self.correct = [0.0 for i in range(cons.trackingFrequency)]
#Run the eLCS algorithm-------------------------------------------------------------------------------
self.run_eLCS()