def prepare_gen_hierarchies_object(dataset, numerical, categorical):
gen_hierarchies_mine = {'categorical': {}, 'range': {}}
# Prepare categorical attributes
for cat_att in categorical:
genh = CGH.CatGenHierarchy(cat_att, config.GENH_DIR + config.GENH_FILE[cat_att])
gen_hierarchies_mine['categorical'][cat_att] = genh
# Prepare numerical attributes
for num_att in numerical:
column = [dataset[idx].get(num_att) for idx in range(len(dataset))]
min_val = min(column)
max_val = max(column)
print("Found " + str(num_att) + " range of: [" + str(min_val) + ":" + str(max_val) + "]")
genh = RGH.RangeGenHierarchy(num_att, min_val, max_val)
gen_hierarchies_mine['range'][num_att] = genh
return gen_hierarchies_mine
def test_catGenRaceGetGenOfWhite(self):
cgh = CGH.CatGenHierarchy('race',
'../data/gen_hierarchies/RaceGH.json')
self.assertEqual(cgh.getGeneralizationOf('White'), 'all')
def test_catGenSexLevels(self):
cgh = CGH.CatGenHierarchy('sex', '../data/gen_hierarchies/SexGH.json')
self.assertEqual(cgh.nrLevels(), 1)
def test_catGenRaceGetLevelEntryOfWhite(self):
cgh = CGH.CatGenHierarchy('race',
'../data/gen_hierarchies/RaceGH.json')
self.assertEqual(cgh.getLevelEntry('White'), 1)
def test_catGenRaceGetNrLevels(self):
cgh = CGH.CatGenHierarchy('race',
'../data/gen_hierarchies/RaceGH.json')
self.assertEqual(cgh.nrLevels(), 1)
def test_catGenCountryGetGenOfAsia(self):
cgh = CGH.CatGenHierarchy(
'native-country', '../data/gen_hierarchies/NativeCountryGH.json')
self.assertEqual(cgh.getGeneralizationOf('Asia'), 'all')
def test_catGenSexGetGenOfMale(self):
cgh = CGH.CatGenHierarchy('sex', '../data/gen_hierarchies/SexGH.json')
self.assertEqual(cgh.getGeneralizationOf('Male'), 'all')
def test_catGenCountryLevels(self):
cgh = CGH.CatGenHierarchy(
'native-country', '../data/gen_hierarchies/NativeCountryGH.json')
self.assertEqual(cgh.nrLevels(), 3)