def generateFirstOrderData(model, numIterations=10000, seqLength=5,
resets=True, suffix='train'):
print "Creating %d iteration file with seqLength %d" % (numIterations, seqLength)
print "Filename",
categoryList, initProbability, transitionTable = model
initProbability = initProbability.cumsum()
transitionTable = transitionTable.cumsum(axis=1)
outputFile = 'fo_%d_%d_%s.csv' % (numIterations, seqLength, suffix)
print "Filename", outputFile
fields = [('reset', 'int', 'R'), ('name', 'string', '')]
o = File(outputFile, fields)
seqIdx = 0
rand = numpy.random.rand()
catIdx = numpy.searchsorted(initProbability, rand)
for i in xrange(numIterations):
rand = numpy.random.rand()
if seqIdx == 0 and resets:
catIdx = numpy.searchsorted(initProbability, rand)
reset = 1
else:
catIdx = numpy.searchsorted(transitionTable[catIdx], rand)
reset = 0
o.write([reset,categoryList[catIdx]])
seqIdx = (seqIdx+1)%seqLength
o.close()
def generateFirstOrderData(model,
numIterations=10000,
seqLength=5,
resets=True,
suffix='train'):
print("Creating %d iteration file with seqLength %d" %
(numIterations, seqLength))
print("Filename", end=' ')
categoryList, initProbability, transitionTable = model
initProbability = initProbability.cumsum()
transitionTable = transitionTable.cumsum(axis=1)
outputFile = 'fo_%d_%d_%s.csv' % (numIterations, seqLength, suffix)
print("Filename", outputFile)
fields = [('reset', 'int', 'R'), ('name', 'string', '')]
o = File(outputFile, fields)
seqIdx = 0
rand = numpy.random.rand()
catIdx = numpy.searchsorted(initProbability, rand)
for i in range(numIterations):
rand = numpy.random.rand()
if seqIdx == 0 and resets:
catIdx = numpy.searchsorted(initProbability, rand)
reset = 1
else:
catIdx = numpy.searchsorted(transitionTable[catIdx], rand)
reset = 0
o.write([reset, categoryList[catIdx]])
seqIdx = (seqIdx + 1) % seqLength
o.close()
def _generateFile(filename, data):
"""
Parameters:
----------------------------------------------------------------
filename: name of .csv file to generate
"""
# Create the file
print "Creating %s..." % (filename)
numRecords, numFields = data.shape
fields = [('field%d'%(i+1), 'float', '') for i in range(numFields)]
outFile = File(filename, fields)
for i in xrange(numRecords):
outFile.write(data[i].tolist())
outFile.close()
