groupby¶

In [5]:

import pandas as pd
import numpy as np
import seaborn as sns

In [2]:

iris = sns.load_dataset("iris")

In [5]:

iris.groupby(iris.species).sum()

Out[5]:

	sepal_length	sepal_width	petal_length	petal_width
species
setosa	250.3	171.4	73.1	12.3
versicolor	296.8	138.5	213.0	66.3
virginica	329.4	148.7	277.6	101.3

In [6]:

def f(df):
    print(df.species.unique())
    return df[df.species != "setosa"].sum()

iris.groupby(iris.species).apply(f)

['setosa']
['setosa']
['versicolor']
['virginica']

Out[6]:

	sepal_length	sepal_width	petal_length	petal_width	species
species
setosa	0.0	0.0	0.0	0.0	0
versicolor	296.8	138.5	213.0	66.3	versicolorversicolorversicolorversicolorversic...
virginica	329.4	148.7	277.6	101.3	virginicavirginicavirginicavirginicavirginicav...

In [11]:

iris.groupby(iris.species).transform(lambda x: print(x.head(2))).head(2)

0    5.1
1    4.9
Name: sepal_length, dtype: float64
0    3.5
1    3.0
Name: sepal_width, dtype: float64
0    1.4
1    1.4
Name: petal_length, dtype: float64
0    0.2
1    0.2
Name: petal_width, dtype: float64
   sepal_length  sepal_width  petal_length  petal_width
0           5.1          3.5           1.4          0.2
1           4.9          3.0           1.4          0.2
50    7.0
51    6.4
Name: sepal_length, dtype: float64
50    3.2
51    3.2
Name: sepal_width, dtype: float64
50    4.7
51    4.5
Name: petal_length, dtype: float64
50    1.4
51    1.5
Name: petal_width, dtype: float64
100    6.3
101    5.8
Name: sepal_length, dtype: float64
100    3.3
101    2.7
Name: sepal_width, dtype: float64
100    6.0
101    5.1
Name: petal_length, dtype: float64
100    2.5
101    1.9
Name: petal_width, dtype: float64

Out[11]:

	sepal_length	sepal_width	petal_length	petal_width
0	NaN	NaN	NaN	NaN
1	NaN	NaN	NaN	NaN

In [13]:

iris.groupby(iris.species)["sepal_length"].transform(lambda x: print(x.head(2))).head(2)

0    5.1
1    4.9
Name: setosa, dtype: float64
50    7.0
51    6.4
Name: versicolor, dtype: float64
100    6.3
101    5.8
Name: virginica, dtype: float64

Out[13]:

0   NaN
1   NaN
Name: sepal_length, dtype: float64

In [16]:

iris.groupby(iris.species).transform(np.sum).iloc[[0,1,50,51,100,101]]

Out[16]:

	sepal_length	sepal_width	petal_length	petal_width
0	250.3	171.4	73.1	12.3
1	250.3	171.4	73.1	12.3
50	296.8	138.5	213.0	66.3
51	296.8	138.5	213.0	66.3
100	329.4	148.7	277.6	101.3
101	329.4	148.7	277.6	101.3

In [17]:

iris.groupby(iris.species).transform(pd.Series.cumsum).iloc[[0,1,50,51,100,101]]

Out[17]:

	sepal_length	sepal_width	petal_length	petal_width
0	5.1	3.5	1.4	0.2
1	10.0	6.5	2.8	0.4
50	7.0	3.2	4.7	1.4
51	13.4	6.4	9.2	2.9
100	6.3	3.3	6.0	2.5
101	12.1	6.0	11.1	4.4

In [18]:

iris.groupby(iris.species)["sepal_length"].transform(pd.Series.cumsum).iloc[[0,1,50,51,100,101]]

Out[18]:

     5.1
    10.0
    7.0
   13.4
   6.3
  12.1
Name: sepal_length, dtype: float64

In [ ]:

In [1]:

import seaborn as sns
planets = sns.load_dataset("planets")
planets.groupby('method')['year'].describe().unstack(level=0)  #.unstack("method")

Out[1]:

method	Astrometry	Eclipse Timing Variations	Imaging	Microlensing	Orbital Brightness Modulation	Pulsar Timing	Pulsation Timing Variations	Radial Velocity	Transit	Transit Timing Variations
count	2.00000	9.000000	38.000000	23.000000	3.000000	5.00000	1.0	553.000000	397.000000	4.000000
mean	2011.50000	2010.000000	2009.131579	2009.782609	2011.666667	1998.40000	2007.0	2007.518987	2011.236776	2012.500000
std	2.12132	1.414214	2.781901	2.859697	1.154701	8.38451	NaN	4.249052	2.077867	1.290994
min	2010.00000	2008.000000	2004.000000	2004.000000	2011.000000	1992.00000	2007.0	1989.000000	2002.000000	2011.000000
25%	2010.75000	2009.000000	2008.000000	2008.000000	2011.000000	1992.00000	2007.0	2005.000000	2010.000000	2011.750000
50%	2011.50000	2010.000000	2009.000000	2010.000000	2011.000000	1994.00000	2007.0	2009.000000	2012.000000	2012.500000
75%	2012.25000	2011.000000	2011.000000	2012.000000	2012.000000	2003.00000	2007.0	2011.000000	2013.000000	2013.250000
max	2013.00000	2012.000000	2013.000000	2013.000000	2013.000000	2011.00000	2007.0	2014.000000	2014.000000	2014.000000

In [2]:

iris = sns.load_dataset("iris")
print(iris.species.unique())
iris.set_index("species").groupby(
    {
        "setosa": "setosa",
        "versicolor": "other",
        "virginica": "other"
    }).mean()

['setosa' 'versicolor' 'virginica']

Out[2]:

	sepal_length	sepal_width	petal_length	petal_width
other	6.262	2.872	4.906	1.676
setosa	5.006	3.428	1.462	0.246

In [3]:

iris.groupby(iris.species == "setosa").mean()

Out[3]:

	sepal_length	sepal_width	petal_length	petal_width
species
False	6.262	2.872	4.906	1.676
True	5.006	3.428	1.462	0.246

In [ ]:

In [6]:

pd.to_timedelta("1days 12:34:56")

Out[6]:

Timedelta('1 days 12:34:56')

In [7]:

dates = pd.DataFrame({
    "da": pd.date_range("2017-01-01", "2017-02-28"),
    "de": pd.timedelta_range("1days 12:34:56.789012345", freq="1H", periods=31+28)
})
dates.info()
dates.groupby(
    [
        pd.Grouper(freq="10d", key="da"),
        dates.da.dt.month,
    ]).count()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 59 entries, 0 to 58
Data columns (total 2 columns):
da    59 non-null datetime64[ns]
de    59 non-null timedelta64[ns]
dtypes: datetime64[ns](1), timedelta64[ns](1)
memory usage: 1.0 KB

Out[7]:

		de
da	da
2017-01-01	1	10
2017-01-11	1	10
2017-01-21	1	10
2017-01-31	1	1
2017-01-31	2	9
2017-02-10	2	10
2017-02-20	2	9

In [8]:

dates.de.dt.components.head()

Out[8]:

	days	hours	minutes	seconds	milliseconds	microseconds	nanoseconds
0	1	12	34	56	789	12	345
1	1	13	34	56	789	12	345
2	1	14	34	56	789	12	345
3	1	15	34	56	789	12	345
4	1	16	34	56	789	12	345

In [ ]: