def test_value_counts_normalize(self):
df_temp = pdc.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 = pdc.DataFrame(
{
"state": np.array(["florida", "texas", "ohio"], dtype=object),
"count": np.array([0.5, 0.375, 0.125]),
}
)
assert_df_equals(df_results[0], df_answer)
df_answer = pdc.DataFrame(
{
"fruit": np.array(["a", "b"], dtype=object),
"count": np.array([0.625, 0.375]),
}
)
assert_df_equals(df_results[1], df_answer)
def test_value_counts(self):
df_temp = pdc.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 = pdc.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 = pdc.DataFrame(
{
"fruit": np.array(["a", "b"], dtype=object),
"count": np.array([5, 3]),
}
)
assert_df_equals(df_results[1], df_answer)
def test_array_length(self):
with pytest.raises(ValueError):
pdc.DataFrame({'a': np.array([1, 2]),
'b': np.array([1])})
# correct construction. no error
pdc.DataFrame({'a': np.array([1, 2]),
'b': np.array([5, 10])})
def test_value_counts(self):
df_temp = pdc.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 = pdc.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 = pdc.DataFrame({
'fruit': np.array(['a', 'b'], dtype=object),
'count': np.array([5, 3])
})
assert_df_equals(df_results[1], df_answer)
with pytest.raises(TypeError):
df_temp.rename(5)
def test_truediv(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_head_tail(self):
df_result = df1.head(2)
df_answer = pdc.DataFrame({'a': a1[:2], 'b': b1[:2], 'c': c1[:2]})
assert_df_equals(df_result, df_answer)
df_result = df1.tail(2)
df_answer = pdc.DataFrame({'a': a1[-2:], 'b': b1[-2:], 'c': c1[-2:]})
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_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_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_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_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_head_tail(self):
df_result = df.head(2)
df_answer = pdc.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 = pdc.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_normalize(self):
df_temp = pdc.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 = pdc.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 = pdc.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 = 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_head_tail(self):
df_result = df.head(2)
df_answer = pdc.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 = pdc.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_new_column(self):
df_result = pdc.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 = pdc.DataFrame({
'a': a,
'b': b,
'c': c,
'd': d,
'e': e,
'f': f
})
assert_df_equals(df_result, df_answer)
def test_value_counts(self):
df_results = df8[['a', 'b']].value_counts()
df_answer = pdc.DataFrame({
'a': np.array(['a', 'b'], dtype=object),
'count': np.array([5, 3])
})
assert_df_equals(df_results[0], df_answer)
df_answer = pdc.DataFrame({
'b': np.array(['A', 'B'], dtype=object),
'count': np.array([4, 4])
})
assert_df_equals(df_results[1], 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)
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_input_types(self):
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_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_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_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_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_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_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_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_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_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)
