def test_one_hot_encode(self):
dataBot = DataBot(self.df, target_name='Survived')
dataBot.impute(['Embarked'], ImputerStrategy.MODE)
dataBot.one_hot_encode('Embarked', self.df['Embarked'].values)
encoded_cols = {'Embarked_S', 'Embarked_C', 'Embarked_Q'}
self.assertTrue(encoded_cols.intersection(set(dataBot.features.columns)) == encoded_cols)
def test_normalize(self):
dataBot = DataBot(self.df, target_name='Survived')
dataBot.impute(['Age'], ImputerStrategy.MEAN)
dataBot.normalize(['Age'])
self.assertTrue(np.allclose(272.84, dataBot.features['Age'].sum(), 2))
def test_impute(self):
dataBot = DataBot(self.df, target_name='Survived')
dataBot.impute(['Age'], ImputerStrategy.MEAN)
expected_a = 21205.17
self.assertTrue(np.allclose(expected_a, dataBot.features['Age'].sum(), 2))