def test_makeTrainingTestingSetBadPercentage(self):
desiredClassStatus = 'Senior'
goodClassStatusDB = GetDatabaseInfoScholarshipsWithClassStatuses(
requirementNeeded=desiredClassStatus)
badClassStatusDB = GetDatabaseInfoScholarshipsWithClassStatuses(
requirementNeeded=desiredClassStatus, useNot=True)
goodClassStatusIds = goodClassStatusDB.getScholarshipsWithClassStatusIdsList(
)
goodClassStatusDataTextList = goodClassStatusDB.getConcatenatedDescriptionsEligibilities(
)
badClassStatusIds = badClassStatusDB.getScholarshipsWithClassStatusIdsList(
)
badClassStatusDataTextList = badClassStatusDB.getConcatenatedDescriptionsEligibilities(
)
trainingPercentage = 2
trainingSetAndTestingSet = MakeDataSet.makeBinaryLabelTrainingAndTestingSet(
firstLabel=desiredClassStatus,
secondLabel='Other',
firstLabelTextList=goodClassStatusDataTextList,
secondLabelTextList=badClassStatusDataTextList,
firstLabelIdsList=goodClassStatusIds,
secondLabelIdsList=badClassStatusIds,
trainingPercentage=trainingPercentage)
self.assertIsNone(trainingSetAndTestingSet)
def __init__(self, classStatus, trainingPercentage, modelSaveFile=None, featuresValuesCountsSaveFile=None):
self.classStatus = classStatus
self.trainingPercentage = trainingPercentage
self.badLabel = 'Other'
self.modelSaveFile = modelSaveFile
self.featuresValueCountsSaveFile = featuresValuesCountsSaveFile
print('Creating datasets...')
goodClassStatusDB = GetDatabaseInfoScholarshipsWithClassStatuses(requirementNeeded=self.classStatus)
badClassStatusDB = GetDatabaseInfoScholarshipsWithClassStatuses(requirementNeeded=self.classStatus, useNot=True)
goodClassStatusIds = goodClassStatusDB.getScholarshipsWithClassStatusIdsList()
goodClassStatusDataTextList = goodClassStatusDB.getConcatenatedDescriptionsEligibilities()
badClassStatusIds = badClassStatusDB.getScholarshipsWithClassStatusIdsList()
badClassStatusDataTextList = badClassStatusDB.getConcatenatedDescriptionsEligibilities()
trainingTestingList = MakeDataSet.makeBinaryLabelTrainingAndTestingSet(
firstLabel=self.classStatus, secondLabel=self.badLabel, firstLabelTextList=goodClassStatusDataTextList,
secondLabelTextList=badClassStatusDataTextList, firstLabelIdsList=goodClassStatusIds,
secondLabelIdsList=badClassStatusIds, trainingPercentage=0.8)
self.training = trainingTestingList[0]
self.testing = trainingTestingList[1]
self.dataFrame = self.makeDataFrame()
self.trainingVectors = []
self.testingVectors = []
self.featuresValueCountsIndexes = []
self.logisticRegressionClassifier = LogisticRegression()
def test_makeTrainingTestingSets(self):
desiredClassStatus = 'Junior'
goodClassStatusDB = GetDatabaseInfoScholarshipsWithClassStatuses(
requirementNeeded=desiredClassStatus)
badClassStatusDB = GetDatabaseInfoScholarshipsWithClassStatuses(
requirementNeeded=desiredClassStatus, useNot=True)
goodClassStatusIds = goodClassStatusDB.getScholarshipsWithClassStatusIdsList(
)
goodClassStatusDataTextList = goodClassStatusDB.getConcatenatedDescriptionsEligibilities(
)
badClassStatusIds = badClassStatusDB.getScholarshipsWithClassStatusIdsList(
)
badClassStatusDataTextList = badClassStatusDB.getConcatenatedDescriptionsEligibilities(
)
trainingPercentage = 0.9
trainingSetAndTestingSet = MakeDataSet.makeBinaryLabelTrainingAndTestingSet(
firstLabel=desiredClassStatus,
secondLabel='Other',
firstLabelTextList=goodClassStatusDataTextList,
secondLabelTextList=badClassStatusDataTextList,
firstLabelIdsList=goodClassStatusIds,
secondLabelIdsList=badClassStatusIds,
trainingPercentage=trainingPercentage)
training = trainingSetAndTestingSet[0]
testing = trainingSetAndTestingSet[1]
totalSetSize = len(training) + len(testing)
self.assertEqual(math.ceil(totalSetSize * trainingPercentage),
len(training))
self.assertEqual((len(goodClassStatusDataTextList) +
len(badClassStatusDataTextList)), totalSetSize)
def test_runEnsembleClassifier(self):
databaseInfo = GetDatabaseInfoScholarshipsWithClassStatuses()
dataTextList = databaseInfo.getConcatenatedDescriptionsEligibilities()
idsList = databaseInfo.getScholarshipsWithClassStatusIdsList()
ensembleClassifyTest = LogisticRegressionMultilabelClassifyClassStatusFromPretrainedModels(dataTextList,
idsList)
ensembleClassifyTest.displayResults()
def test_makeTrainingTestingSetBadPercentage(self):
desiredClassStatus = 'Senior'
goodClassStatusDB = GetDatabaseInfoScholarshipsWithClassStatuses(requirementNeeded=desiredClassStatus)
badClassStatusDB = GetDatabaseInfoScholarshipsWithClassStatuses(requirementNeeded=desiredClassStatus,
useNot=True)
goodClassStatusIds = goodClassStatusDB.getScholarshipsWithClassStatusIdsList()
goodClassStatusDataTextList = goodClassStatusDB.getConcatenatedDescriptionsEligibilities()
badClassStatusIds = badClassStatusDB.getScholarshipsWithClassStatusIdsList()
badClassStatusDataTextList = badClassStatusDB.getConcatenatedDescriptionsEligibilities()
trainingPercentage = 2
trainingSetAndTestingSet = MakeDataSet.makeBinaryLabelTrainingAndTestingSet(
firstLabel=desiredClassStatus, secondLabel='Other', firstLabelTextList=goodClassStatusDataTextList,
secondLabelTextList=badClassStatusDataTextList, firstLabelIdsList=goodClassStatusIds,
secondLabelIdsList=badClassStatusIds, trainingPercentage=trainingPercentage)
self.assertIsNone(trainingSetAndTestingSet)
def __init__(self,
classStatus,
trainingPercentage,
modelSaveFile=None,
featuresValuesCountsSaveFile=None):
self.classStatus = classStatus
self.trainingPercentage = trainingPercentage
self.badLabel = 'Other'
self.modelSaveFile = modelSaveFile
self.featuresValueCountsSaveFile = featuresValuesCountsSaveFile
print('Creating datasets...')
goodClassStatusDB = GetDatabaseInfoScholarshipsWithClassStatuses(
requirementNeeded=self.classStatus)
badClassStatusDB = GetDatabaseInfoScholarshipsWithClassStatuses(
requirementNeeded=self.classStatus, useNot=True)
goodClassStatusIds = goodClassStatusDB.getScholarshipsWithClassStatusIdsList(
)
goodClassStatusDataTextList = goodClassStatusDB.getConcatenatedDescriptionsEligibilities(
)
badClassStatusIds = badClassStatusDB.getScholarshipsWithClassStatusIdsList(
)
badClassStatusDataTextList = badClassStatusDB.getConcatenatedDescriptionsEligibilities(
)
trainingTestingList = MakeDataSet.makeBinaryLabelTrainingAndTestingSet(
firstLabel=self.classStatus,
secondLabel=self.badLabel,
firstLabelTextList=goodClassStatusDataTextList,
secondLabelTextList=badClassStatusDataTextList,
firstLabelIdsList=goodClassStatusIds,
secondLabelIdsList=badClassStatusIds,
trainingPercentage=0.8)
self.training = trainingTestingList[0]
self.testing = trainingTestingList[1]
self.dataFrame = self.makeDataFrame()
self.trainingVectors = []
self.testingVectors = []
self.featuresValueCountsIndexes = []
self.logisticRegressionClassifier = LogisticRegression()
def test_makeTrainingTestingSets(self):
desiredClassStatus = 'Junior'
goodClassStatusDB = GetDatabaseInfoScholarshipsWithClassStatuses(requirementNeeded=desiredClassStatus)
badClassStatusDB = GetDatabaseInfoScholarshipsWithClassStatuses(requirementNeeded=desiredClassStatus,
useNot=True)
goodClassStatusIds = goodClassStatusDB.getScholarshipsWithClassStatusIdsList()
goodClassStatusDataTextList = goodClassStatusDB.getConcatenatedDescriptionsEligibilities()
badClassStatusIds = badClassStatusDB.getScholarshipsWithClassStatusIdsList()
badClassStatusDataTextList = badClassStatusDB.getConcatenatedDescriptionsEligibilities()
trainingPercentage = 0.9
trainingSetAndTestingSet = MakeDataSet.makeBinaryLabelTrainingAndTestingSet(
firstLabel=desiredClassStatus, secondLabel='Other', firstLabelTextList=goodClassStatusDataTextList,
secondLabelTextList=badClassStatusDataTextList, firstLabelIdsList=goodClassStatusIds,
secondLabelIdsList=badClassStatusIds, trainingPercentage=trainingPercentage)
training = trainingSetAndTestingSet[0]
testing = trainingSetAndTestingSet[1]
totalSetSize = len(training) + len(testing)
self.assertEqual(math.ceil(totalSetSize * trainingPercentage), len(training))
self.assertEqual((len(goodClassStatusDataTextList) + len(badClassStatusDataTextList)), totalSetSize)
def test_runModelOnFreshman(self):
classStatus = 'Freshman'
modelSaveFile = '..\ClassifierTrainedModels\%sClassStatusTrainedLRModel' % classStatus
featuresValueCountsSaveFile = '..\ClassifierTrainedModels\%sClassStatusTrainedFeaturesValueCounts' % classStatus
# test to make sure can open model and fvc save files:
testModelSaveOpen = open(modelSaveFile, 'rb')
testModelSaveOpen.close()
testFVCSaveOpen = open(featuresValueCountsSaveFile, 'rb')
testFVCSaveOpen.close()
# now do test
databaseInfo = GetDatabaseInfoScholarshipsWithClassStatuses()
dataTextList = databaseInfo.getConcatenatedDescriptionsEligibilities()
idsList = databaseInfo.getScholarshipsWithClassStatusIdsList()
testClassify = LogisticRegressionClassifyClassStatusFromPretrainedModel(trainedModelInputFile=modelSaveFile,
trainedFeaturesValueCountsIndexesFile=featuresValueCountsSaveFile,
testingDataTextList=dataTextList,
testingDataIdsList=idsList)
testClassify.displayResults()
def test_RunEnsembleClassifierInsertResultsIntoDB(self):
# get data from db
databaseInfoExtract = GetDatabaseInfoScholarshipsWithClassStatuses()
dataTextList = databaseInfoExtract.getConcatenatedDescriptionsEligibilities()
idsList = databaseInfoExtract.getScholarshipsWithClassStatusIdsList()
# run classifier, return results
ensembleClassifyTest = LogisticRegressionMultilabelClassifyClassStatusFromPretrainedModels(dataTextList,
idsList)
idsAndPredictionsList = ensembleClassifyTest.doAllClassificationsAndReturnFilteredPredictionsListsById()
# insert results into db
db = SUDBConnect()
ids = [prediction[0] for prediction in idsAndPredictionsList]
predictions = [prediction[1] for prediction in idsAndPredictionsList]
for scholarshipWithClassStatusID, prediction in zip(ids, predictions):
prediction = ', '.join(prediction)
db.insertUpdateOrDeleteDB(
"update dbo.ScholarshipsWithClassStatuses set EnsembleClassifierPrediction='" + prediction + "' where ScholarshipsWithClassStatusId = '" + str(
scholarshipWithClassStatusID) + "'")
def test_runModelOnFreshman(self):
classStatus = 'Freshman'
modelSaveFile = '..\ClassifierTrainedModels\%sClassStatusTrainedLRModel' % classStatus
featuresValueCountsSaveFile = '..\ClassifierTrainedModels\%sClassStatusTrainedFeaturesValueCounts' % classStatus
# test to make sure can open model and fvc save files:
testModelSaveOpen = open(modelSaveFile, 'rb')
testModelSaveOpen.close()
testFVCSaveOpen = open(featuresValueCountsSaveFile, 'rb')
testFVCSaveOpen.close()
# now do test
databaseInfo = GetDatabaseInfoScholarshipsWithClassStatuses()
dataTextList = databaseInfo.getConcatenatedDescriptionsEligibilities()
idsList = databaseInfo.getScholarshipsWithClassStatusIdsList()
testClassify = LogisticRegressionClassifyClassStatusFromPretrainedModel(
trainedModelInputFile=modelSaveFile,
trainedFeaturesValueCountsIndexesFile=featuresValueCountsSaveFile,
testingDataTextList=dataTextList,
testingDataIdsList=idsList)
testClassify.displayResults()