def mapSequences(rows):
# triage into three separate lists
# transition, emission, and gold
goldTests = []
for row in rows:
personRecords = row[1]
sorted(row[1], key=lambda x: (x[6], x[2]))
startState = "START"
previous = startState
keepInModel = not utils.isTest()
goldStandard = []
for item in personRecords:
# handle transition
current = buildTransitionWrapper(item[5], item[4], item[2])
# handle emission
emissionKey = previous + "_" + current
patientAmount = item[3]
# keep in model?
#new dictionaries - yield creates each one
if (keepInModel):
yield ((previous, current, transitionType), 1)
yield ((emissionKey, patientAmount, emissionType), 1)
else:
goldStandard.append((previous, current, emissionKey, patientAmount, item[4], item[5]))
previous = current
# push gold standard, if it exists
if not keepInModel:
yield ((row[0][0], row[0][1], goldType), goldStandard)
def determineDictionary(self, test, train): if utils.isTest(self.setSplit): return (test, True) return (train, False)