def testValidateSalesData_allSuccesses(self):
df = getDfFromFile(self.INVALID_EXCEL_ALL_SUCCESSES)
featuresList, featureNameMapping = getFeaturesFromDf(df)
expectedErrorMessages = [
Messages.INVALID_SUCCESS_VALUES]
# Test
validSalesData, errorMessages = validateSalesData(featuresList,
featureNameMapping,
df)
# Validate
self.assertEqual(validSalesData, False)
self.assertEqual(errorMessages, expectedErrorMessages)
def testValidateSalesData_allProblems(self):
df = getDfFromFile(self.INVALID_CSV_ALL_PROBLEMS)
featuresList, featureNameMapping = getFeaturesFromDf(df)
expectedErrorMessages = [
Messages.MISSING_CLIENT_ID_COLUMN,
Messages.INVALID_SUCCESS_VALUES,
Messages.createEmptyCellsInColumnsMessage(['Referral', 'Opportunity Business Area'])]
# Test
validSalesData, errorMessages = validateSalesData(featuresList,
featureNameMapping,
df)
# Validate
self.assertEqual(validSalesData, False)
self.assertEqual(errorMessages, expectedErrorMessages)
def testValidateSalesData_missingRequiredCols(self):
df = getDfFromFile(self.INVALID_CSV_MISSING_REQUIRED_COLS)
featuresList, featureNameMapping = getFeaturesFromDf(df)
expectedErrorMessages = [
Messages.MISSING_CLIENT_ID_COLUMN,
Messages.MISSING_SALESPERSON_ID_COLUMN,
Messages.MISSING_SUCCESS_COLUMN]
# Test
validSalesData, errorMessages = validateSalesData(featuresList,
featureNameMapping,
df)
# Validate
self.assertEqual(validSalesData, False)
self.assertEqual(errorMessages, expectedErrorMessages)
def testValidatePreprocessingInput_invalidNumericaText(self):
df = getDfFromFile(self.INVALID_EXCEL_INVALID_NUMERICAL_TEXT_FEATURE)
formField1 = PreprocessingFormField('NumericalNumber', FeatureTypeChoice.NUMERICAL.value, DataTypeChoice.NUMBER.value)
formField2 = PreprocessingFormField('ValidNumericalText', FeatureTypeChoice.NUMERICAL.value, DataTypeChoice.TEXT.value)
formField3 = PreprocessingFormField('InvalidNumericalText', FeatureTypeChoice.NUMERICAL.value, DataTypeChoice.TEXT.value)
formFields = [formField1, formField2, formField3]
expectedErrorMessages = [
Messages.createInvalidNumericalFeatureTypeMessage(
formField3.featureName, formField3.featureType, formField3.dataType)]
# Test
validPreprocessingInput, errorMessages = validatePreprocessingInput(formFields, df)
# Validate
self.assertEqual(validPreprocessingInput, False)
def testGetFeaturesFromDf(self):
df = getDfFromFile(self.INVALID_CSV_ALL_PROBLEMS)
expectedFeatureNameMapping = {
'client ids': 'Client IDs',
'referral': 'Referral',
'sales process': 'Sales Process',
'opportunity business area': 'Opportunity Business Area',
'salesperson id': 'Salesperson ID',
'success': 'Success',
}
expectedInputFeatures = ['client ids', 'referral', 'sales process',
'opportunity business area', 'salesperson id', 'success']
# Test
featuresList, featureNameMapping = getFeaturesFromDf(df)
# Validate
self.assertEqual(featuresList, expectedInputFeatures)
self.assertEqual(featureNameMapping, expectedFeatureNameMapping)
def testValidatePreprocessingInput_invalidBooleanNumber(self):
df = getDfFromFile(self.INVALID_EXCEL_INVALID_BOOLEAN_NUMBER_FEATURE)
formField1 = PreprocessingFormField('ValidBooleanNumber', FeatureTypeChoice.BOOLEAN.value, DataTypeChoice.NUMBER.value)
formField2 = PreprocessingFormField('ValidBooleanNumber2', FeatureTypeChoice.BOOLEAN.value, DataTypeChoice.NUMBER.value)
formField3 = PreprocessingFormField('ValidBooleanNumber3', FeatureTypeChoice.BOOLEAN.value, DataTypeChoice.NUMBER.value)
formField4 = PreprocessingFormField('InvalidBooleanNumber', FeatureTypeChoice.BOOLEAN.value, DataTypeChoice.NUMBER.value)
formFields = [formField1, formField2, formField3, formField4]
expectedErrorMessages = [
Messages.createInvalidBooleanFeatureTypeMessage(
formField4.featureName, formField4.featureType, formField4.dataType),
]
# Test
validPreprocessingInput, errorMessages = validatePreprocessingInput(formFields, df)
# Validate
self.assertEqual(validPreprocessingInput, False)
self.assertEqual(errorMessages, expectedErrorMessages)