reframe

In [1]:

# https://dplyr.tidyverse.org/reference/summarise.html
%run nb_helpers.py

from datar.data import starwars
from datar.all import *

nb_header(reframe)

# https://dplyr.tidyverse.org/reference/summarise.html %run nb_helpers.py from datar.data import starwars from datar.all import * nb_header(reframe)

Try this notebook on binder.

★ reframe

¶

Reframe a data frame.¶

See original API
https://dplyr.tidyverse.org/reference/reframe.html

Args:¶

  _data: A data frame
  *args: and
  **kwargs: Name-value pairs, where value is the reframed
    data for each group

Returns:¶

  A data frame with the reframed columns

In [2]:

table = c("a", "b", "d", "f")

df = tibble(
  g = c(1, 1, 1, 2, 2, 2, 2),
  x = c("e", "a", "b", "c", "f", "d", "a")
)

# `reframe()` allows you to apply functions that return
# an arbitrary number of rows
df >> reframe(x = intersect(f.x, table))

table = c("a", "b", "d", "f") df = tibble( g = c(1, 1, 1, 2, 2, 2, 2), x = c("e", "a", "b", "c", "f", "d", "a") ) # `reframe()` allows you to apply functions that return # an arbitrary number of rows df >> reframe(x = intersect(f.x, table))

Out[2]:

	x
	<object>
0	a
1	b
2	f
3	d

In [3]:

df >> group_by(f.g)

df >> group_by(f.g)

Out[3]:

	g	x
	<int64>	<object>
0	1	e
1	1	a
2	1	b
3	2	c
4	2	f
5	2	d
6	2	a

TibbleGrouped: g (n=2)

In [4]:

# The output is always ungrouped, even when using `group_by()`
(
  df
    >> group_by(f.g)
    >> reframe(x = intersect(f.x, table))
)

# The output is always ungrouped, even when using `group_by()` ( df >> group_by(f.g) >> reframe(x = intersect(f.x, table)) )

Out[4]:

	g	x
	<Int64>	<string>
0	1	a
1	1	b
2	2	f
3	2	d
4	2	a

In [12]:

from pandas.core.groupby import SeriesGroupBy
from pipda import register_func

@register_func
def quantile_df(x, probs=[0.25, 0.5, 0.75]):
    return tibble(
        val = quantile(x, probs, na_rm=True),
        quant = [probs] if isinstance(x, SeriesGroupBy) else probs
    )


x = [10, 15, 18, 12]
quantile_df(x)

from pandas.core.groupby import SeriesGroupBy from pipda import register_func @register_func def quantile_df(x, probs=[0.25, 0.5, 0.75]): return tibble( val = quantile(x, probs, na_rm=True), quant = [probs] if isinstance(x, SeriesGroupBy) else probs ) x = [10, 15, 18, 12] quantile_df(x)

Out[12]:

	val	quant
	<float64>	<float64>
0	11.50	0.25
1	13.50	0.50
2	15.75	0.75

In [13]:

starwars >> reframe(quantile_df(f.height))

starwars >> reframe(quantile_df(f.height))

Out[13]:

	val	quant
	<float64>	<float64>
0	167.0	0.25
1	180.0	0.50
2	191.0	0.75

In [14]:

(
    starwars
    >> group_by(f.homeworld)
    >> reframe(quantile_df(f.height))
)

( starwars >> group_by(f.homeworld) >> reframe(quantile_df(f.height)) )

Out[14]:

	homeworld	val	quant
	<string>	<Float64>	<Float64>
0	Tatooine	165.5	0.25
1	Tatooine	175.0	0.5
2	Tatooine	183.0	0.75
3	Naboo	165.0	0.25
...	...	...	...
4	Naboo	183.0	0.5
142	Umbara	178.0	0.5
143	Umbara	178.0	0.75
144	Utapau	206.0	0.25
145	Utapau	206.0	0.5
146	Utapau	206.0	0.75

147 rows × 3 columns

In [ ]:

(
    starwars
    >> group_by(f.homeworld)
    >> reframe(
        across(c(f.height, f.mass), quantile_df)
    )
)

( starwars >> group_by(f.homeworld) >> reframe( across(c(f.height, f.mass), quantile_df) ) )

/home/pwwang/.cache/pypoetry/virtualenvs/datar-TA_GutPO-py3.12/lib/python3.12/site-packages/numpy/lib/_nanfunctions_impl.py:1598: RuntimeWarning: All-NaN slice encountered
  return _nanquantile_unchecked(

Out[ ]:

	homeworld	height$val	height$quant	mass$val	mass$quant
	<string>	<Float64>	<Float64>	<Float64>	<Float64>
0	Tatooine	165.5	0.25	75.0	0.25
1	Tatooine	175.0	0.5	80.5	0.5
2	Tatooine	183.0	0.75	93.0	0.75
3	Naboo	165.0	0.25	50.25	0.25
...	...	...	...	...	...
4	Naboo	183.0	0.5	70.5	0.5
142	Umbara	178.0	0.5	48.0	0.5
143	Umbara	178.0	0.75	48.0	0.75
144	Utapau	206.0	0.25	80.0	0.25
145	Utapau	206.0	0.5	80.0	0.5
146	Utapau	206.0	0.75	80.0	0.75

147 rows × 5 columns