def anti_join(self,
y,
on=None,
on_x=None,
on_y=None,
sort=True,
suffix_y="_y"):
tf = tidyframe(self._obj, copy=False, check=False)
y = tidyframe(y, copy=False, check=True)
return tf.anti_join(y,
on=on,
on_x=on_x,
on_y=on_y,
sort=sort,
suffix_y=suffix_y).to_pandas(copy=False)
def test_summarise(penguins_data):
penguins_tidy = tidyframe(penguins_data, copy=False)
expected = pd.DataFrame({
"A": [
np.mean(penguins_data["bill_length_mm"] +
penguins_data["bill_depth_mm"])
],
"B": [
np.mean(penguins_data["bill_length_mm"] -
penguins_data["bill_depth_mm"])
]
})
result = penguins_tidy.summarise({
('A', 'B'): ("[np.mean(x + y), np.mean(x - y)]",
["bill_length_mm", "bill_depth_mm"])
}).to_pandas()
assert_frame_equal_v2(expected, result)
# summarise across
expected = pd.DataFrame({
"bill_length_mm": [np.mean(penguins_data["bill_length_mm"])],
"bill_depth_mm": [np.mean(penguins_data["bill_depth_mm"])]
})
result = penguins_tidy.summarise(
func=np.mean, column_names=['bill_length_mm',
'bill_depth_mm']).to_pandas()
assert_frame_equal_v2(expected, result)
def test_mutate_across(penguins_data):
penguins_tidy = tidyframe(penguins_data, copy=False)
# across mode with column names
expected = penguins_data[['bill_length_mm', 'body_mass_g']]\
.assign(demean_bill_length_mm=lambda x: x["bill_length_mm"] - np.mean(x["bill_length_mm"]))\
.assign(demean_body_mass_g=lambda x: x["body_mass_g"] - np.mean(x["body_mass_g"]))
result = (penguins_tidy.select(['bill_length_mm', 'body_mass_g']).mutate(
column_names=['bill_length_mm', 'body_mass_g'],
func=lambda x: x - np.mean(x),
prefix="demean_")).to_pandas()
assert_frame_equal_v2(expected, result)
# grouped across with column names
expected = penguins_data[['bill_length_mm', 'body_mass_g', 'species']]
expected = expected.groupby("species")\
.apply(lambda x: x.assign(demean_bill_length_mm=lambda y: y["bill_length_mm"] - np.mean(y["bill_length_mm"]))\
.assign(demean_body_mass_g=lambda y: y["body_mass_g"] - np.mean(y["body_mass_g"])))
result = (penguins_tidy.select(
['bill_length_mm', 'body_mass_g',
'species']).mutate(column_names=['bill_length_mm', 'body_mass_g'],
func=lambda x: x - np.mean(x),
prefix="demean_",
by='species')).to_pandas()
assert_frame_equal_v2(expected, result)
def test_mutate_groupby(penguins_data):
penguins_tidy = tidyframe(penguins_data, copy=False)
expected = penguins_data.groupby("sex", dropna=False)\
.apply(lambda x: x.assign(year_mod=lambda y: y["year"] + np.mean(y["year"]) - 4015))
result = penguins_tidy.mutate(
{
'year_mod': "x['year'] + np.mean(x['year']) - 4015"
}, by='sex').to_pandas()
assert_frame_equal_v2(expected, result)
def separate(self,
column_name,
into,
sep='[^[:alnum:]]+',
strict=True,
keep=False):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.separate(column_name=column_name,
into=into,
sep=sep,
strict=strict,
keep=keep).to_pandas(copy=False)
def fill_na(self,
column_direction_dict,
order_by=None,
ascending=True,
na_position="last",
by=None):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.fill_na(column_direction_dict=column_direction_dict,
order_by=order_by,
ascending=ascending,
na_position=na_position,
by=by).to_pandas(copy=False)
def pivot_longer(self,
cols,
names_to="name",
values_to="value",
include=True,
values_drop_na=False):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.pivot_longer(
cols=cols,
names_to=names_to,
values_to=values_to,
include=include,
values_drop_na=values_drop_na).to_pandas(copy=False)
def test_summarise_groupby(penguins_data):
penguins_tidy = tidyframe(penguins_data, copy=False)
# expected = penguins_data.groupby(["species", "sex"], dropna=False)\
# .apply(lambda x: x[["year"]].mean().rename({"year": "a_mean"}))\
# .reset_index().sort_values(["species", "sex"])
expected = penguins_data.groupby(["species", "sex"], dropna=False)\
.apply(lambda x: np.mean(x["year"]))\
.reset_index().rename(columns={0: "a_mean"})\
.sort_values(["species", "sex"]).reset_index(drop=True)
expected = expected.convert_dtypes()
result = penguins_tidy.summarise({"a_mean": (np.mean, 'year')},
by = ['species', 'sex']
).to_pandas()\
.sort_values(["species", "sex"]).reset_index(drop=True)
assert_frame_equal_v2(expected, result)
# groupby summarise with kwargs
expected = penguins_data.groupby(["species", "sex"], dropna=False)\
.apply(lambda x: np.mean(x["year"]) + 4)\
.reset_index().rename(columns={0: "a_mean"})\
.sort_values(["species", "sex"]).reset_index(drop=True)
expected = expected.convert_dtypes()
result = penguins_tidy.summarise(
{
"a_mean": lambda x, **kwargs: x['year'].mean() + kwargs['leap']
},
by=['species', 'sex'],
leap=4).to_pandas().sort_values(["species",
"sex"]).reset_index(drop=True)
assert_frame_equal_v2(expected, result)
# groupby summarise across
expected = penguins_data.groupby(["species", "sex"], dropna=False)\
.apply(lambda x: x.loc[:, x.columns[x.apply(types.is_numeric_dtype, axis=0)]].mean()\
.add_prefix("avg_"))\
.reset_index()\
.sort_values(["species", "sex"]).reset_index(drop=True)
expected = expected.convert_dtypes()
result = penguins_tidy.summarise(func=np.mean,
predicate=types.is_numeric_dtype,
prefix="avg_",
by=['species',
'sex']).to_pandas().sort_values(
["species",
"sex"]).reset_index(drop=True)
assert_frame_equal_v2(expected, result)
def slice_max(self,
n=None,
prop=None,
order_by_column=None,
with_ties=True,
rounding_type="round",
by=None):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.slice_max(n=n,
prop=prop,
order_by_column=order_by_column,
with_ties=with_ties,
rounding_type=rounding_type,
by=by).to_pandas(copy=False)
def slice_sample(self,
n=None,
prop=None,
replace=False,
weights=None,
random_state=None,
by=None):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.slice_sample(n=n,
prop=prop,
replace=replace,
weights=weights,
random_state=random_state,
by=by).to_pandas(copy=False)
def pivot_wider(self,
names_from,
values_from,
values_fill=None,
values_fn=None,
id_cols=None,
sep="__"):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.pivot_wider(names_from=names_from,
values_from=values_from,
values_fill=values_fill,
values_fn=values_fn,
id_cols=id_cols,
sep=sep).to_pandas(copy=False)
def group_modify(self,
func,
by,
preserve_row_order=False,
row_order_column_name="rowid_temp",
is_pandas_udf=False,
**kwargs):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.group_modify(func=func,
by=by,
preserve_row_order=preserve_row_order,
row_order_column_name=row_order_column_name,
is_pandas_udf=is_pandas_udf,
**kwargs).to_pandas(copy=False)
def test_mutate_orderby(penguins_data):
penguins_tidy = tidyframe(penguins_data, copy=False)
expected = (penguins_data[['year', 'species', 'bill_length_mm']]\
.sort_values("bill_length_mm")\
.assign(year_cumsum=lambda x: np.cumsum(x["year"]))\
.sort_index())
result = (penguins_tidy.select(['year', 'species',
'bill_length_mm']).mutate(
{
'year_cumsum': (np.cumsum, 'year')
},
order_by='bill_length_mm').to_pandas())
assert_frame_equal_v2(expected, result)
def summarise(self,
dictionary=None,
func=None,
column_names=None,
predicate=None,
prefix="",
by=None,
**kwargs):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.summarise(dictionary=dictionary,
func=func,
column_names=column_names,
predicate=predicate,
prefix=prefix,
by=by,
**kwargs).to_pandas(copy=False)
def test_select(penguins_data):
penguins_tidy = tidyframe(penguins_data, copy=False)
sel_cols = penguins_data.convert_dtypes().columns[
penguins_data.convert_dtypes().apply(lambda x: x.dtype != "string",
axis=0)]
expected = penguins_data.loc[:, sel_cols]
result = penguins_tidy.select(
predicate=lambda x: x.dtype != "string").to_pandas()
assert_frame_equal_v2(expected, result)
sel_cols = list(
set(penguins_data.columns).difference(set(['sex', 'species'])))
expected = penguins_data.loc[:, sel_cols]
result = penguins_tidy.select(['sex', 'species'],
include=False).to_pandas()
assert_frame_equal_v2(expected, result)
def separate_rows(self, column_name, sep=';'):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.separate_rows(column_name=column_name,
sep=sep).to_pandas(copy=False)
def add_group_number(self, by=None, name='group_number'):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.add_group_number(by=by, name=name).to_pandas(copy=False)
def add_row_number(self, name='row_number', by=None):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.add_row_number(name=name, by=by).to_pandas(copy=False)
def colnames(self):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.colnames
def dim(self):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.dim
def shape(self):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.shape
def split(self, by):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.split(by=by).to_pandas(copy=False)
def unnest(self, nest_column_name='data'):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.unnest(nest_column_name=nest_column_name).to_pandas(
copy=False)
def ncol(self):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.ncol
def nest(self, column_names=None, nest_column_name='data', include=True):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.nest(column_names=column_names,
nest_column_name=nest_column_name,
include=include).to_pandas(copy=False)
def nest_by(self, by=None, nest_column_name='data', drop_by=True):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.nest_by(by=by,
nest_column_name=nest_column_name,
drop_by=drop_by).to_pandas(copy=False)
def select(self, column_names=None, predicate=None, include=True):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.select(column_names=column_names,
predicate=predicate,
include=include).to_pandas(copy=False)
def drop_na(self, column_names=None):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.drop_na(column_names=column_names).to_pandas(copy=False)
def relocate(self, column_names, before=None, after=None):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.relocate(column_names=column_names,
before=before,
after=after).to_pandas(copy=False)
def unite(self, column_names, into, sep="_", keep=False):
tf = tidyframe(self._obj, copy=False, check=False)
return tf.unite(column_names=column_names,
into=into,
sep=sep,
keep=keep).to_pandas(copy=False)
