def test_gt_lt(self):
df_result = df5 > 3
df_answer = pdc.DataFrame({'a': a5 > 3, 'b': b5 > 3})
assert_df_equals(df_result, df_answer)
df_result = df5 < 2
df_answer = pdc.DataFrame({'a': a5 < 2, 'b': b5 < 2})
assert_df_equals(df_result, df_answer)
def test_ge_le(self):
df_result = df5 >= 3
df_answer = pdc.DataFrame({'a': a5 >= 3, 'b': b5 >= 3})
assert_df_equals(df_result, df_answer)
df_result = df5 < 2
df_answer = pdc.DataFrame({'a': a5 <= 2, 'b': b5 <= 2})
assert_df_equals(df_result, df_answer)
def test_sub(self):
df_result = df5 - 3
df_answer = pdc.DataFrame({'a': a5 - 3, 'b': b5 - 3})
assert_df_equals(df_result, df_answer)
df_result = 3 - df5
df_answer = pdc.DataFrame({'a': 3 - a5, 'b': 3 - b5})
assert_df_equals(df_result, df_answer)
def test_eq_ne(self):
df_result = df5 == 3
df_answer = pdc.DataFrame({'a': a5 == 3, 'b': b5 == 3})
assert_df_equals(df_result, df_answer)
df_result = df5 != 2
df_answer = pdc.DataFrame({'a': a5 != 2, 'b': b5 != 2})
assert_df_equals(df_result, df_answer)
def test_pow(self):
df_result = df5**3
df_answer = pdc.DataFrame({'a': a5**3, 'b': b5**3})
assert_df_equals(df_result, df_answer)
df_result = 2**df5
df_answer = pdc.DataFrame({'a': 2**a5, 'b': 2**b5})
assert_df_equals(df_result, df_answer)
def test_floordiv(self):
df_result = df5 // 3
df_answer = pdc.DataFrame({'a': a5 // 3, 'b': b5 // 3})
assert_df_equals(df_result, df_answer)
df_result = 3 // df5
df_answer = pdc.DataFrame({'a': 3 // a5, 'b': 3 // b5})
assert_df_equals(df_result, df_answer)
def test_input_types(self): ## test _check_input_types
with pytest.raises(TypeError):
pdc.DataFrame([1, 2, 3])
with pytest.raises(TypeError):
pdc.DataFrame({1: 5, 'b': 10})
with pytest.raises(TypeError):
pdc.DataFrame({'a': np.array([1]), 'b': 10})
with pytest.raises(ValueError):
pdc.DataFrame({'a': np.array([1]), 'b': np.array([[1]])})
# correct construction. no error
pdc.DataFrame({'a': np.array([1]), 'b': np.array([1])})
def test_std(self):
df_result = df1.std()
df_answer = pdc.DataFrame({
'b': np.array([b1.std()]),
'c': np.array([np.nan])
})
assert_df_equals(df_result, df_answer)
def test_median(self):
df_result = df1.median()
df_answer = pdc.DataFrame({
'b': np.array([8]),
'c': np.array([np.nan])
})
assert_df_equals(df_result, df_answer)
def test_clip(self):
df_result = df42.clip(0, 4)
df_answer = pdc.DataFrame({
'a': np.array([0, 4, 3]),
'b': np.array([3.4, 4, 0])
})
assert_df_equals(df_result, df_answer)
def test_mul(self):
df_result = df5 * 3
df_answer = pdc.DataFrame({'a': a5 * 3, 'b': b5 * 3})
assert_df_equals(df_result, df_answer)
df_result = 3 * df5
assert_df_equals(df_result, df_answer)
def test_add(self):
df_result = df5 + 3
df_answer = pdc.DataFrame({'a': a5 + 3, 'b': b5 + 3})
assert_df_equals(df_result, df_answer)
df_result = 3 + df5
assert_df_equals(df_result, df_answer)
def test_diff(self):
df_result = df42.diff(1)
df_answer = pdc.DataFrame({
'a': np.array([np.nan, 16, -2]),
'b': np.array([np.nan, 1.7, -11.1])
})
assert_df_equals(df_result, df_answer)
def test_cumsum(self):
df_result = df42.cumsum()
df_answer = pdc.DataFrame({
'a': np.array([-11, -6, -3]),
'b': np.array([3.4, 8.5, 2.5])
})
assert_df_equals(df_result, df_answer)
def test_round(self):
df_result = df42.round(0)
df_answer = pdc.DataFrame({
'a': np.array([-11, 5, 3]),
'b': np.array([3, 5, -6])
})
assert_df_equals(df_result, df_answer)
def test_cummax(self):
df_result = df42.cummax()
df_answer = pdc.DataFrame({
'a': np.array([-11, 5, 5]),
'b': np.array([3.4, 5.1, 5.1])
})
assert_df_equals(df_result, df_answer)
def test_cummin(self):
df_result = df42.cummin()
df_answer = pdc.DataFrame({
'a': np.array([-11, -11, -11]),
'b': np.array([3.4, 3.4, -6])
})
assert_df_equals(df_result, df_answer)
def test_nunique(self):
df_result = df4.nunique()
df_answer = pdc.DataFrame({
'a': np.array([2]),
'b': np.array([2]),
'c': np.array([2])
})
assert_df_equals(df_result, df_answer)
def test_sum(self):
df_result = df1.sum()
df_answer = pdc.DataFrame({
'a': np.array(['abc'], dtype='O'),
'b': np.array([24]),
'c': np.array([np.nan])
})
assert_df_equals(df_result, df_answer)
def test_any(self):
df_result = df2.any()
df_answer = pdc.DataFrame({
'a': np.array([True]),
'b': np.array([True]),
'c': np.array([True])
})
assert_df_equals(df_result, df_answer)
def test_argmin(self):
df_result = df1.argmin()
df_answer = pdc.DataFrame({
'a': np.array([0]),
'b': np.array([1]),
'c': np.array([1])
})
assert_df_equals(df_result, df_answer)
def test_isna(self):
df_result = df3.isna()
df_answer = pdc.DataFrame({
'a': np.array([False, True, False]),
'b': np.array([False, False, False]),
'c': np.array([False, True, False])
})
assert_df_equals(df_result, df_answer)
def test_count(self):
df_result = df3.count()
df_answer = pdc.DataFrame({
'a': np.array([2]),
'b': np.array([3]),
'c': np.array([2])
})
assert_df_equals(df_result, df_answer)
def test_dtypes(self):
cols = np.array(['a', 'b', 'c', 'd', 'e'], dtype='O')
dtypes = np.array(['string', 'string', 'float', 'bool', 'int'],
dtype='O')
df_result = df.dtypes
df_answer = pdc.DataFrame({'Column Name': cols, 'Data Type': dtypes})
assert_df_equals(df_result, df_answer)
def test_col_slice(self):
df_answer = pdc.DataFrame({'a': a, 'b': b, 'c': c})
assert_df_equals(df[:, :3], df_answer)
df_answer = pdc.DataFrame({'a': a[::2], 'b': b[::2], 'c': c[::2]})
assert_df_equals(df[::2, :3], df_answer)
df_answer = pdc.DataFrame({
'a': a[::2],
'b': b[::2],
'c': c[::2],
'd': d[::2],
'e': e[::2]
})
assert_df_equals(df[::2, :], df_answer)
with pytest.raises(TypeError):
df[:, set()]
def test_pct_change(self):
df_result = df42.pct_change(1)
df_answer = pdc.DataFrame({
'a':
np.array([np.nan, 16 / -11, -2 / 5]),
'b':
np.array([np.nan, 1.7 / 3.4, -11.1 / 5.1])
})
assert_df_equals(df_result, df_answer)
def test_simple_boolean(self):
bool_arr = np.array([True, False, False])
df_bool = pdc.DataFrame({'col': bool_arr})
df_result = df[df_bool]
df_answer = pdc.DataFrame({
'a': a[bool_arr],
'b': b[bool_arr],
'c': c[bool_arr],
'd': d[bool_arr],
'e': e[bool_arr]
})
assert_df_equals(df_result, df_answer)
with pytest.raises(ValueError):
df_bool = pdc.DataFrame({'col': bool_arr, 'col2': bool_arr})
df[df_bool]
with pytest.raises(TypeError):
df_bool = pdc.DataFrame({'col': np.array[1, 2, 3]})
def test_list_columns(self):
df_answer = pdc.DataFrame({'c': c, 'e': e})
assert_df_equals(df[:, [2, 4]], df_answer)
assert_df_equals(df[:, [2, 'e']], df_answer)
assert_df_equals(df[:, ['c', 'e']], df_answer)
df_result = df[2, ['a', 'e']]
df_answer = pdc.DataFrame({'a': a[[2]], 'e': e[[2]]})
assert_df_equals(df_result, df_answer)
df_answer = pdc.DataFrame({'c': c[[1, 2]], 'e': e[[1, 2]]})
assert_df_equals(df[[1, 2], ['c', 'e']], df_answer)
df1 = pdc.DataFrame({
'a': np.array([True, False, True]),
'b': np.array([1, 3, 5])
})
df_answer = pdc.DataFrame({'c': c[[0, 2]], 'e': e[[0, 2]]})
assert_df_equals(df[df1['a'], ['c', 'e']], df_answer)
def test_pivot_table_both(self):
df_result = df8.pivot_table(rows='a',
columns='b',
values='c',
aggfunc='sum')
df_answer = pdc.DataFrame({
'a': np.array(['a', 'b'], dtype=object),
'A': np.array([9., 8.]),
'B': np.array([13., 6.])
})
assert_df_equals(df_result, df_answer)
def test_sample(self):
df_result = df7.sample(2, seed=1)
df_answer = pdc.DataFrame({
'a': np.array(['a', 'a'], dtype=object),
'b': np.array([2., 5.1])
})
assert_df_equals(df_result, df_answer)
df_result = df7.sample(frac=.7, seed=1)
df_answer = pdc.DataFrame({
'a': np.array(['a', 'a', 'b'], dtype=object),
'b': np.array([2., 5.1, 6.])
})
assert_df_equals(df_result, df_answer)
with pytest.raises(TypeError):
df7.sample(2.5)
with pytest.raises(ValueError):
df7.sample(frac=-2)