def splitDataFrame(self, dataSet, sensitiveVal):
sensitiveAttr = dataSet.protectedAttribute
df = dataSet.trainDataFrame
ds = DataSet()
ds.fileName = dataSet.fileName
ds.protectedAttribute = sensitiveAttr
ds.trueLabels = dataSet.trueLabels
ds.trainHeaders = dataSet.trainHeaders
ds.numAttributes = dataSet.numAttributes
try:
ds.trainDataFrame = df.groupby([sensitiveAttr
]).get_group(sensitiveVal)
except:
return 0
return ds
def determineGroups(self, dataSet):
df = dataSet.testDataFrame
possibleGroups = df[dataSet.protectedAttribute].unique()
organizedDataSetList = []
for value in possibleGroups:
# Setting up the group as a new DataSet
newDataSet = DataSet()
newDataSet.fileName = dataSet.fileName
newDataSet.testDataFrame = df[df[dataSet.protectedAttribute] ==
value]
newDataSet.protectedAttribute = dataSet.protectedAttribute
newDataSet.trueLabels = dataSet.trueLabels
newDataSet.headers = dataSet.headers
newDataSet.testHeaders = dataSet.testHeaders
newDataSet.numAttributes = dataSet.numAttributes
#resets indices for later indexing
newDataSet.testDataFrame.reset_index(inplace=True, drop=True)
organizedDataSetList.append(newDataSet)
return organizedDataSetList, possibleGroups
