def __init__(self):
databaseInfo = GetDatabaseInfoScholarshipsWithClassStatuses()
self.actualLabels = databaseInfo.getRequirementNeededList()
self.predictedLabels = databaseInfo.getEnsembleClassifierPredictions()
self.actualLabels = self.convertActualLabelStringsToList()
self.predictedLabels = self.convertPredictedLabelStringsToList()
self.calculateExactMatchAccuracy()
self.calculateNumLabelsAccuracy()
print("Average Accuracy Across Labels: %.2f percent" %
self.calculateAverageAccuracyWithinLists(self.actualLabels,
self.predictedLabels))
self.calculateListAccuracyWithinMatchedGreaterOrLesserListLengths()
def test_CheckLabelsListFormat(self):
# get the list of labels, 100
dbinfo = GetDatabaseInfoScholarshipsWithClassStatuses()
listOLabels = dbinfo.getRequirementNeededList()
self.assertEqual(type(listOLabels), list)
self.assertEqual(type(listOLabels[1]), list)
first100Labels = listOLabels[:100]
self.assertEqual(len(first100Labels), 100)
# get fake data
X, Y = make_multilabel_classification(n_classes=10, n_labels=3, allow_unlabeled=False)
# test the OVR
testClassifiier = OneVsRestClassifier(LogisticRegression())
testClassifiier.fit(X, first100Labels)