def test_normalRun(self):
# set up
db = SUDBConnect()
sponsorsList = []
descriptionList = []
ocList = []
iefaLeadTrainingIdList = []
actualBad = []
concatenatedDescriptionOCList = []
rows = db.getRowsDB("select * from dbo.IefaLeadsTrainingItems where BadScholarship!='Maybe'")
for row in rows:
sponsorsList.append(row.Sponsor)
descriptionList.append(row.Description)
ocList.append(row.OtherCriteria)
actualBad.append(row.BadScholarship)
iefaLeadTrainingIdList.append(str(row.IefaLeadTrainingId))
for i in range(len(descriptionList)):
conatenatedItem = '%s %s' % (descriptionList[i], ocList[i])
concatenatedDescriptionOCList.append(conatenatedItem)
# test
testNER = ClassifyBadScholarships(sponsorsList, concatenatedDescriptionOCList)
predictedBad = testNER.loopThroughLeadsAndDoStuff()
accuracy = ComputeAccuracy(actualBad, predictedBad).calculateAccuracy()
print(accuracy)
# put result in db
for i in range(len(predictedBad)):
iefaLeadTrainingId = iefaLeadTrainingIdList[i]
predicted = predictedBad[i]
db.insertUpdateOrDeleteDB(
"update dbo.IefaLeadsTrainingItems set NormalTestPredictedTag='" + predicted + "' where IefaLeadTrainingId='" + iefaLeadTrainingId + "'")
def test_infoTextOnlyNoInsert(self):
# set up
db = SUDBConnect()
sponsorsList = []
descriptionList = []
ocList = []
iefaLeadTrainingIdList = []
actualBad = []
concatenatedDescriptionOCList = []
rows = db.getRowsDB("select * from dbo.IefaLeadsTrainingItems where BadScholarship!='Maybe'")
for row in rows:
sponsorsList.append(row.Sponsor)
descriptionList.append(row.Description)
ocList.append(row.OtherCriteria)
actualBad.append(row.BadScholarship)
iefaLeadTrainingIdList.append(str(row.IefaLeadTrainingId))
for i in range(len(descriptionList)):
conatenatedItem = '%s %s' % (descriptionList[i], ocList[i])
concatenatedDescriptionOCList.append(conatenatedItem)
# test
testNER = ClassifyBadScholarships(sponsorsList, concatenatedDescriptionOCList, test='infoTextOnly')
infoTextPredictedBad = testNER.loopThroughLeadsAndDoStuff()
accuracy = ComputeAccuracy(actualBad, infoTextPredictedBad).calculateAccuracy()
print(accuracy)
def checkBadScholarship(leadArray, fundingClassification):
if fundingClassification == 'Scholarship':
sponsor = leadArray[4]
infoText = '%s %s %s' % (leadArray[3], leadArray[6], leadArray[7])
badScholarshipClassifier = ClassifyBadScholarships()
badScholarshipPrediction = badScholarshipClassifier.classifyOpportunity(sponsor, infoText)
return badScholarshipPrediction
else:
return ''
def checkBadScholarship(leadArray, fundingClassification):
if fundingClassification == 'Scholarship':
sponsor = leadArray[4]
infoText = '%s %s' % (leadArray[7], leadArray[8])
badScholarshipClassifier = ClassifyBadScholarships()
badScholarshipPrediction = badScholarshipClassifier.classifyOpportunity(sponsor, infoText)
return badScholarshipPrediction
else:
return ''
def test_checkSponsorBadInstitutions(self):
nertest = ClassifyBadScholarships(['blah'], ['blah'])
sponsor1 = 'University of Arizona'
testSponsor1 = nertest.checkBadSponsor(sponsor1)
self.assertFalse(testSponsor1)
sponsor2 = 'University of Colorado'
testSponsor2 = nertest.checkBadSponsor(sponsor2)
self.assertTrue(testSponsor2)
sponsor3 = 'Chicken and Egg Group'
testSponsor3 = nertest.checkBadSponsor(sponsor3)
self.assertFalse(testSponsor3)
sponsor4 = 'School of Bacon'
testSponsor4 = nertest.checkBadSponsor(sponsor4)
self.assertTrue(testSponsor4)
def getPredictedBad(self):
predictedBad = ClassifyBadScholarships(
self.sponsorList, self.infoTextList).loopThroughLeadsAndDoStuff()
return predictedBad
def test_CheckBadTextRewrite(self):
nertest = ClassifyBadScholarships(['blah'], ['blah'])
infoText = 'CollegeWeekLive gives away $1,000 per month just for viewing and visiting US colleges online. Winning is easy - all you need to do is login to CollegeWeekLive and visit 3 colleges that interest you. One lucky winner will be awarded a $1,000 scholarship every month. And many other winners may find the college of their dreams.'
testText = nertest.checkBadText(infoText)
def test_CheckBadText(self):
nertest = ClassifyBadScholarships(['blah'], ['blah'])
infoText = 'I go to the University of Arizona in Tucson, Arizona'
testText = nertest.checkBadText(infoText)
self.assertFalse(testText)