def test_array_length(self):
with pytest.raises(ValueError):
mpd.DataFrame({'a': np.array([1, 2]),
'b': np.array([1])})
# correct construction. no error
mpd.DataFrame({'a': np.array([1, 2]),
'b': np.array([5, 10])})
def test_pow(self):
df_result = df5 ** 3
df_answer = mpd.DataFrame({'a': a5 ** 3, 'b': b5 ** 3})
assert_df_equals(df_result, df_answer)
df_result = 2 ** df5
df_answer = mpd.DataFrame({'a': 2 ** a5, 'b': 2 ** b5})
assert_df_equals(df_result, df_answer)
def test_sub(self):
df_result = df5 - 3
df_answer = mpd.DataFrame({'a': a5 - 3, 'b': b5 - 3})
assert_df_equals(df_result, df_answer)
df_result = 3 - df5
df_answer = mpd.DataFrame({'a': 3 - a5, 'b': 3 - b5})
assert_df_equals(df_result, df_answer)
def test_gt_lt(self):
df_result = df5 > 3
df_answer = mpd.DataFrame({'a': a5 > 3, 'b': b5 > 3})
assert_df_equals(df_result, df_answer)
df_result = df5 < 2
df_answer = mpd.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 = mpd.DataFrame({'a': a5 >= 3, 'b': b5 >= 3})
assert_df_equals(df_result, df_answer)
df_result = df5 < 2
df_answer = mpd.DataFrame({'a': a5 <= 2, 'b': b5 <= 2})
assert_df_equals(df_result, df_answer)
def test_floordiv(self):
df_result = df5 // 3
df_answer = mpd.DataFrame({'a': a5 // 3, 'b': b5 // 3})
assert_df_equals(df_result, df_answer)
df_result = 3 // df5
df_answer = mpd.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 = mpd.DataFrame({'a': a5 == 3, 'b': b5 == 3})
assert_df_equals(df_result, df_answer)
df_result = df5 != 2
df_answer = mpd.DataFrame({'a': a5 != 2, 'b': b5 != 2})
assert_df_equals(df_result, df_answer)
def test_head_tail(self):
df_result = df.head(2)
df_answer = mpd.DataFrame({'a': a[:2], 'b': b[:2], 'c': c[:2],
'd': d[:2], 'e': e[:2]})
assert_df_equals(df_result, df_answer)
df_result = df.tail(2)
df_answer = mpd.DataFrame({'a': a[-2:], 'b': b[-2:], 'c': c[-2:],
'd':d[-2:], 'e': e[-2:]})
assert_df_equals(df_result, df_answer)
def test_value_counts(self):
df_temp = mpd.DataFrame({'state': np.array(['texas', 'texas', 'texas', 'florida', 'florida', 'florida', 'florida', 'ohio']),
'fruit': np.array(['a', 'a', 'a', 'a', 'b', 'b', 'b', 'a'])})
df_results = df_temp.value_counts()
df_answer = mpd.DataFrame({'state': np.array(['florida', 'texas', 'ohio'], dtype=object),
'count': np.array([4, 3, 1])})
assert_df_equals(df_results[0], df_answer)
df_answer = mpd.DataFrame({'fruit': np.array(['a', 'b'], dtype=object),
'count': np.array([5, 3])})
assert_df_equals(df_results[1], df_answer)
def test_value_counts_normalize(self):
df_temp = mpd.DataFrame({'state': np.array(['texas', 'texas', 'texas', 'florida', 'florida', 'florida', 'florida', 'ohio']),
'fruit': np.array(['a', 'a', 'a', 'a', 'b', 'b', 'b', 'a'])})
df_results = df_temp.value_counts(normalize=True)
df_answer = mpd.DataFrame({'state': np.array(['florida', 'texas', 'ohio'], dtype=object),
'count': np.array([.5, .375, .125])})
assert_df_equals(df_results[0], df_answer)
df_answer = mpd.DataFrame({'fruit': np.array(['a', 'b'], dtype=object),
'count': np.array([.625, .375])})
assert_df_equals(df_results[1], df_answer)
def test_col_slice(self):
df_answer = mpd.DataFrame({'a': a, 'b': b, 'c': c})
assert_df_equals(df[:, :3], df_answer)
df_answer = mpd.DataFrame({'a': a[::2], 'b': b[::2], 'c': c[::2]})
assert_df_equals(df[::2, :3], df_answer)
df_answer = mpd.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_simple_boolean(self):
bool_arr = np.array([True, False, False])
df_bool = mpd.DataFrame({'col': bool_arr})
df_result = df[df_bool]
df_answer = mpd.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 = mpd.DataFrame({'col': bool_arr, 'col2': bool_arr})
df[df_bool]
with pytest.raises(TypeError):
df_bool = mpd.DataFrame({'col': np.array[1, 2, 3]})
def test_sample(self):
df_result = df7.sample(2, seed=1)
df_answer = mpd.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 = mpd.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)
def test_add(self):
df_result = df5 + 3
df_answer = mpd.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_data_types(self):
df_result = df_emp.dtypes
cols = np.array(['dept', 'race', 'gender', 'salary'], dtype='O')
dtypes = np.array(['string', 'string', 'string', 'int'], dtype='O')
df_answer = mpd.DataFrame({'Column Name': cols,
'Data Type': dtypes})
assert_df_equals(df_result, df_answer)
def test_mul(self):
df_result = df5 * 3
df_answer = mpd.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_input_types(self):
with pytest.raises(TypeError):
mpd.DataFrame([1, 2, 3])
with pytest.raises(TypeError):
mpd.DataFrame({1: 5, 'b': 10})
with pytest.raises(TypeError):
mpd.DataFrame({'a': np.array([1]), 'b': 10})
with pytest.raises(ValueError):
mpd.DataFrame({'a': np.array([1]),
'b': np.array([[1]])})
# correct construction. no error
mpd.DataFrame({'a': np.array([1]),
'b': np.array([1])})
def test_list_columns(self):
df_answer = mpd.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 = mpd.DataFrame({'a': a[[2]], 'e': e[[2]]})
assert_df_equals(df_result, df_answer)
df_answer = mpd.DataFrame({'c': c[[1, 2]], 'e': e[[1, 2]]})
assert_df_equals(df[[1, 2], ['c', 'e']], df_answer)
df1 = mpd.DataFrame({'a': np.array([True, False, True]),
'b': np.array([1, 3, 5])})
df_answer = mpd.DataFrame({'c': c[[0, 2]], 'e': e[[0, 2]]})
assert_df_equals(df[df1['a'], ['c', 'e']], 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 = mpd.DataFrame({'Column Name': cols,
'Data Type': dtypes})
assert_df_equals(df_result, df_answer)
def test_all_row_selections(self):
df1 = mpd.DataFrame({'a': np.array([True, False, True]),
'b': np.array([1, 3, 5])})
with pytest.raises(ValueError):
df[df1, 'e']
with pytest.raises(TypeError):
df[df1['b'], 'c']
df_result = df[df1['a'], 'c']
df_answer = mpd.DataFrame({'c': c[[True, False, True]]})
assert_df_equals(df_result, df_answer)
df_result = df[[1, 2], 0]
df_answer = mpd.DataFrame({'a': a[[1, 2]]})
assert_df_equals(df_result, df_answer)
df_result = df[1:, 0]
assert_df_equals(df_result, df_answer)
def test_pivot_table_rows_or_cols(self):
df_result = df8.pivot_table(rows='a')
df_answer = mpd.DataFrame({'a': np.array(['a', 'b'], dtype=object),
'size': np.array([5, 3])})
assert_df_equals(df_result, df_answer)
df_result = df8.pivot_table(rows='a', values='c', aggfunc='sum')
df_answer = mpd.DataFrame({'a': np.array(['a', 'b'], dtype=object),
'sum': np.array([22, 14])})
assert_df_equals(df_result, df_answer)
df_result = df8.pivot_table(columns='b')
df_answer = mpd.DataFrame({'A': np.array([4]),
'B': np.array([4])})
assert_df_equals(df_result, df_answer)
df_result = df8.pivot_table(columns='a', values='c', aggfunc='sum')
df_answer = mpd.DataFrame({'a': np.array([22]), 'b': np.array([14])})
assert_df_equals(df_result, df_answer)
def test_head(self):
data = {'dept': np.array(['Houston Police Department-HPD',
'Houston Fire Department (HFD)',
'Houston Police Department-HPD',
'Public Works & Engineering-PWE',
'Houston Airport System (HAS)'], dtype='O'),
'race': np.array(['White', 'White', 'Black', 'Asian', 'White'], dtype='O'),
'gender': np.array(['Male', 'Male', 'Male', 'Male', 'Male'], dtype='O'),
'salary': np.array([45279, 63166, 66614, 71680, 42390])}
result = df_emp.head()
answer = mpd.DataFrame(data)
assert_df_equals(result, answer)
def test_new_column(self):
df_result = mpd.DataFrame({'a': a, 'b': b, 'c': c, 'd': d, 'e': e})
f = np.array([1.5, 23, 4.11])
df_result['f'] = f
df_answer = mpd.DataFrame({'a': a, 'b': b, 'c': c, 'd': d, 'e': e, 'f': f})
assert_df_equals(df_result, df_answer)
df_result = mpd.DataFrame({'a': a, 'b': b, 'c': c, 'd': d, 'e': e})
df_result['f'] = True
f = np.repeat(True, 3)
df_answer = mpd.DataFrame({'a': a, 'b': b, 'c': c, 'd': d, 'e': e, 'f': f})
assert_df_equals(df_result, df_answer)
df_result = mpd.DataFrame({'a': a, 'b': b, 'c': c, 'd': d, 'e': e})
f = np.array([1.5, 23, 4.11])
df_result['c'] = f
df_answer = mpd.DataFrame({'a': a, 'b': b, 'c': f, 'd': d, 'e': e})
assert_df_equals(df_result, df_answer)
with pytest.raises(NotImplementedError):
df[['a', 'b']] = 5
with pytest.raises(ValueError):
df['a'] = np.random.rand(5, 5)
with pytest.raises(ValueError):
df['a'] = np.random.rand(5)
with pytest.raises(ValueError):
df['a'] = df[['a', 'b']]
with pytest.raises(ValueError):
df1 = mpd.DataFrame({'a': np.random.rand(5)})
df['a'] = df1
with pytest.raises(TypeError):
df['a'] = set()
def test_istitle(self):
result = df_string.str.istitle('num')
num = np.array([False, False])
answer = mpd.DataFrame({'num': num})
assert_df_equals(result, answer)
def test_zfill(self):
result = df_string.str.zfill('movie', 16)
movie = np.array(['0field of dreams', '0000000star wars'], dtype='O')
answer = mpd.DataFrame({'movie': movie})
assert_df_equals(result, answer)
def test_upper(self):
result = df_string.str.upper('movie')
movie = np.array(['FIELD OF DREAMS', 'STAR WARS'], dtype='O')
answer = mpd.DataFrame({'movie': movie})
assert_df_equals(result, answer)
def test_title(self):
result = df_string.str.title('movie')
movie = np.array(['Field Of Dreams', 'Star Wars'], dtype='O')
answer = mpd.DataFrame({'movie': movie})
assert_df_equals(result, answer)
def test_replace(self):
result = df_string.str.replace('movie', 's', 'Z')
movie = np.array(['field of dreamZ', 'Ztar warZ'], dtype='O')
answer = mpd.DataFrame({'movie': movie})
assert_df_equals(result, answer)
def test_strip(self):
result = df_string.str.strip('movie', 'fs')
movie = np.array(['ield of dream', 'tar war'], dtype='O')
answer = mpd.DataFrame({'movie': movie})
assert_df_equals(result, answer)
def test_isupper(self):
result = df_string.str.isupper('movie')
movie = np.array([False, False])
answer = mpd.DataFrame({'movie': movie})
assert_df_equals(result, answer)