def test_delete_single_column_number(self):
result = render(
pd.DataFrame({
'A': [0, 0, 1],
'B': ['monkey', 'kangaroo', 'cat']
}), P(['A'], 'delete'))
expected = pd.DataFrame({'A': [0, 1], 'B': ['monkey', 'cat']})
assert_frame_equal(result, expected)
def test_delete_single_column_string(self):
result = render(
pd.DataFrame({
'A': ['apple', 'apple', 'orange'],
'B': ['monkey', 'kangaroo', 'cat'],
}), P(['A'], 'delete'))
expected = pd.DataFrame({
'A': ['apple', 'orange'],
'B': ['monkey', 'cat']
})
assert_frame_equal(result, expected)
def test_cumcount_single_column_number(self):
result = render(
pd.DataFrame({
'A': [0, 0, 1],
'B': ['monkey', 'kangaroo', 'cat']
}), P(['A'], 'cumcount'))
expected = pd.DataFrame({
'A': [0, 0, 1],
'B': ['monkey', 'kangaroo', 'cat'],
'Duplicate number': [1, 2, 1]
})
assert_frame_equal(result, expected)
def test_delete_unused_categories(self):
result = render(
pd.DataFrame({
'A': ['a', 'b', 'c'],
'B': ['d', 'd', 'e']
},
dtype='category'), P(['B'], 'delete'))
expected = pd.DataFrame({
'A': ['a', 'c'],
'B': ['d', 'e']
},
dtype='category')
assert_frame_equal(result, expected)
def test_cumcount_multi_column(self):
result = render(
pd.DataFrame({
'A': ['apple', 'orange', 'apple'],
'B': [0, 'kangaroo', 0],
'C': ['monkey', 'kangaroo', 'cat'],
}), P(['A', 'B'], 'cumcount'))
expected = pd.DataFrame({
'A': ['apple', 'orange', 'apple'],
'B': [0, 'kangaroo', 0],
'C': ['monkey', 'kangaroo', 'cat'],
'Duplicate number': [1, 1, 2]
})
assert_frame_equal(result, expected)
def test_no_colnames(self):
# No colnames -> do nothing
result = render(pd.DataFrame({'A': ['', np.nan, 'x']}), P([],
'delete'))
expected = pd.DataFrame({'A': ['', np.nan, 'x']})
assert_frame_equal(result, expected)