def TestDiscretize():
data = mat2csv()
sc = SymbolicClustering(segmentLength=10,
paaSize=2,
alphabetSize=10,
upperBound=3000,
lowerBound=0)
for i in range(0, 10):
segments = sc.discretize(data[i][0][0:90, 6])
sc.printSegments()
def TestGrammarInduction():
data = mat2csv()
sc = SymbolicClustering(segmentLength=10,
paaSize=2,
alphabetSize=10,
upperBound=3000,
lowerBound=0)
for i in range(0, 10):
segments = sc.discretize(data[i][0][0:90, 6])
#sc.printSegments()
sc.grammar_induction(segments)
print(sc.grammar.print_grammar())
print('--------------------------------------')
def TestClustersGeneration():
data = mat2csv()
sc = SymbolicClustering(segmentLength=10,
paaSize=2,
alphabetSize=10,
upperBound=3000,
lowerBound=0)
for i in range(0, 10):
segments = sc.discretize(data[i][0][0:90, 6])
sc.grammar_induction(segments)
frequencyMatrix = sc.get_frequency_matrix()
windows = sc.cut_window(frequencyMatrix)
sc.generateInitialClusters(0, windows)
for window in windows:
window.printClusters()
def TestCutWindow():
data = mat2csv()
sc = SymbolicClustering(segmentLength=10,
paaSize=2,
alphabetSize=10,
upperBound=3000,
lowerBound=0)
for i in range(0, 10):
segments = sc.discretize(data[i][0][0:90, 6])
sc.grammar_induction(segments)
frequencyMatrix = sc.get_frequency_matrix()
windows = sc.cut_window(frequencyMatrix)
print('--------------------------------------')
for window in windows:
print(str(window.startIndex) + ',' + str(window.endIndex))
def TestGetFrequencyMatrix():
data = mat2csv()
sc = SymbolicClustering(segmentLength=10,
paaSize=2,
alphabetSize=10,
upperBound=3000,
lowerBound=0)
for i in range(0, 10):
segments = sc.discretize(data[i][0][0:90, 6])
sc.grammar_induction(segments)
print(sc.grammar.print_grammar())
frequencyMatrix = sc.get_frequency_matrix()
print(frequencyMatrix)
print('--------------------------------------')
def patternGenerate(self):
"""
@description :
generate patterns from the timeseries, where we will use the symbolic clustering method
to generate windows in each dimension. The pattern is defined as the center of clusters
from each window.
---------
@param :
-------
@Returns : listPatterns
-------
"""
print('正在生成patterns...')
numDim = len(self.dataSet)
listPatterns = []
for i in range(0, numDim):
print(i / numDim)
timeSeries = self.dataSet[i]
seriesNum = len(timeSeries)
if self.pStep > seriesNum:
raise PStepIsNotValid()
sc = SymbolicClustering(segmentLength=self.segmentLength,
paaSize=self.paaSize,
alphabetSize=self.alphabetSize,
upperBound=self.listBounds[i][1],
lowerBound=self.listBounds[i][0])
for j in range(0, self.pStep):
# print(timeSeries[j])
segments = sc.discretize(timeSeries[j])
sc.grammar_induction(segments)
frequencyMatrix = sc.get_frequency_matrix()
windows = sc.cut_window(frequencyMatrix)
windows = sc.generateInitialClusters(0, windows)
# for window in windows:
# window.printClustersInDetail()
dynamicStep = self.pStep
sc = SymbolicClustering(segmentLength=self.segmentLength,
paaSize=self.paaSize,
alphabetSize=self.alphabetSize,
upperBound=self.listBounds[i][1],
lowerBound=self.listBounds[i][0])
while dynamicStep < seriesNum:
if len(timeSeries[dynamicStep]) < self.seriesLen:
# for window in windows:
# window.printClustersInDetail()
timeSeries[dynamicStep] = timeSeries[dynamicStep].tolist()
for j in range(len(timeSeries[dynamicStep]),
self.seriesLen):
timeSeries[dynamicStep].append(0.0)
print(timeSeries[dynamicStep])
DynamicClustering.dynamicClustering(timeSeries[dynamicStep],
windows, dynamicStep,
self.segmentLength)
segments = sc.discretize(timeSeries[dynamicStep])
sc.grammar_induction(segments)
if (dynamicStep + 1) % self.pStep == 0:
frequencyMatrix = sc.get_frequency_matrix()
scwindows = sc.cut_window(frequencyMatrix)
# scwindows = sc.generateInitialClusters(dynamicStep + 1 - self.pStep, scwindows)
windows = ParametersSelfTuning.windowSplitting(
windows, scwindows)
windows = ParametersSelfTuning.windowCombination(windows)
sc = SymbolicClustering(segmentLength=self.segmentLength,
paaSize=self.paaSize,
alphabetSize=self.alphabetSize,
upperBound=self.listBounds[i][1],
lowerBound=self.listBounds[i][0])
dynamicStep += 1
listPatterns.append(windows)
return listPatterns
def TestWindowCombination():
data = mat2csv()
sc = SymbolicClustering(segmentLength=10,
paaSize=2,
alphabetSize=10,
upperBound=3000,
lowerBound=0)
for i in range(0, 10):
segments = sc.discretize(data[i][0][0:90, 6])
sc.grammar_induction(segments)
frequencyMatrix = sc.get_frequency_matrix()
windows = sc.cut_window(frequencyMatrix)
windows = sc.generateInitialClusters(0, windows)
print('--------0-10 by now:-------------')
for window in windows:
window.printClusters()
sc = SymbolicClustering(segmentLength=10,
paaSize=2,
alphabetSize=10,
upperBound=3000,
lowerBound=0)
for i in range(10, 20):
segments = sc.discretize(data[i][0][0:90, 6])
sc.grammar_induction(segments)
DynamicClustering.dynamicClustering(data[i][0][0:90, 6], windows, i,
10)
frequencyMatrix = sc.get_frequency_matrix()
scwindows = sc.cut_window(frequencyMatrix)
scwindows = sc.generateInitialClusters(10, scwindows)
print('--------10-20 by sc:-------------')
for window in scwindows:
window.printClusters()
print('--------10-20 by now:-------------')
for window in windows:
window.printClusters()
windows = ParametersSelfTuning.windowSplitting(windows, scwindows)
print('--------10-20 by now split:-------------')
for window in windows:
window.printClusters()
windows = ParametersSelfTuning.windowCombination(windows)
print('--------10-20 by now combination:-------------')
for window in windows:
window.printClustersInDetail()
print('--------0-20 original:-------------')
for i in range(0, 20):
print(data[i][0][0:90, 6])