聚合运算1

import pandas as pd
import numpy as np

df = pd.DataFrame({"key1":["a","a","b","b","a"],
                  "key2":["one","two","one","two","one"],
                  "data1":np.random.randint(1,10,5),
                  "data2":np.random.randint(1,10,5)})
df

	data1	data2	key1	key2
0	9	5	a	one
1	2	6	a	two
2	2	9	b	one
3	7	4	b	two
4	4	7	a	one

df.groupby("key1").sum()

	data1	data2
key1
a	15	18
b	9	13

df.groupby("key1").describe()

		data1	data2
key1
a	count	3.000000	3.000000
	mean	5.000000	6.000000
	std	3.605551	1.000000
	min	2.000000	5.000000
	25%	3.000000	5.500000
	50%	4.000000	6.000000
	75%	6.500000	6.500000
	max	9.000000	7.000000
b	count	2.000000	2.000000
	mean	4.500000	6.500000
	std	3.535534	3.535534
	min	2.000000	4.000000
	25%	3.250000	5.250000
	50%	4.500000	6.500000
	75%	5.750000	7.750000
	max	7.000000	9.000000

grouped = df.groupby("key1")

def peak_range(s):
    print(type(s))
    return s.max() - s.min()
grouped.agg(peak_range)

<class 'pandas.core.series.Series'>
<class 'pandas.core.series.Series'>
<class 'pandas.core.series.Series'>
<class 'pandas.core.series.Series'>
<class 'pandas.core.series.Series'>
<class 'pandas.core.series.Series'>
<class 'pandas.core.series.Series'>

	data1	data2
key1
a	7	2
b	5	5

grouped.agg(["std","mean","sum",peak_range])

<class 'pandas.core.series.Series'>
<class 'pandas.core.series.Series'>
<class 'pandas.core.series.Series'>
<class 'pandas.core.series.Series'>

	data1				data2
	std	mean	sum	peak_range	std	mean	sum	peak_range
key1
a	3.605551	5.0	15	7	1.000000	6.0	18	2
b	3.535534	4.5	9	5	3.535534	6.5	13	5

grouped.agg(["std","mean","sum",("range", peak_range)])

<class 'pandas.core.series.Series'>
<class 'pandas.core.series.Series'>
<class 'pandas.core.series.Series'>
<class 'pandas.core.series.Series'>

	data1				data2
	std	mean	sum	range	std	mean	sum	range
key1
a	3.605551	5.0	15	7	1.000000	6.0	18	2
b	3.535534	4.5	9	5	3.535534	6.5	13	5

d = {"data1":"mean",
    "data2":"sum"}
grouped.agg(d)

	data1	data2
key1
a	5.0	18
b	4.5	13

d = {"data1":["mean",("range", peak_range)],
    "data2":"sum"}
grouped.agg(d)

<class 'pandas.core.series.Series'>
<class 'pandas.core.series.Series'>

	data1		data2
	mean	range	sum
key1
a	5.0	7	18
b	4.5	5	13

grouped.agg(d).reset_index()

<class 'pandas.core.series.Series'>
<class 'pandas.core.series.Series'>

	key1	data1		data2
		mean	range	sum
0	a	5.0	7	18
1	b	4.5	5	13

df = pd.DataFrame({"key1":["a","a","b","b","a"],
                  "key2":["one","two","one","two","one"],
                  "data1":np.random.randint(1,10,5),
                  "data2":np.random.randint(1,10,5)})
df

	data1	data2	key1	key2
0	6	2	a	one
1	1	4	a	two
2	1	7	b	one
3	6	4	b	two
4	4	7	a	one

kl_mean = df.groupby("key1").mean().add_prefix("mean_")

kl_mean

	mean_data1	mean_data2
key1
a	3.666667	4.333333
b	3.500000	5.500000

pd.merge(df,kl_mean, left_on="key1", right_index=True)

	data1	data2	key1	key2	mean_data1	mean_data2
0	6	2	a	one	3.666667	4.333333
1	1	4	a	two	3.666667	4.333333
4	4	7	a	one	3.666667	4.333333
2	1	7	b	one	3.500000	5.500000
3	6	4	b	two	3.500000	5.500000

kl_mean = df.groupby("key1").transform(np.mean).add_prefix("mean_")
kl_mean

	mean_data1	mean_data2
0	3.666667	4.333333
1	3.666667	4.333333
2	3.500000	5.500000
3	3.500000	5.500000
4	3.666667	4.333333

df[kl_mean.columns] = kl_mean
df

	data1	data2	key1	key2	mean_data1	mean_data2
0	6	2	a	one	3.666667	4.333333
1	1	4	a	two	3.666667	4.333333
2	1	7	b	one	3.500000	5.500000
3	6	4	b	two	3.500000	5.500000
4	4	7	a	one	3.666667	4.333333

df = pd.DataFrame(np.random.randint(1,10,(5,5)),
                 columns=["a","b","c","d","e"],
                 index=["alice","bob","candy","dark","emily"])
df

	a	b	c	d	e
alice	9	8	5	4	2
bob	2	8	3	3	6
candy	9	4	4	4	8
dark	7	6	8	1	7
emily	2	4	5	7	1

def demean(s):
    return s - s.mean()
key = ["one","one","two","one","two"]
demeaned = df.groupby(key).transform(demean)
demeaned

	a	b	c	d	e
alice	3.0	0.666667	-0.333333	1.333333	-3.0
bob	-4.0	0.666667	-2.333333	0.333333	1.0
candy	3.5	0.000000	-0.500000	-1.500000	3.5
dark	1.0	-1.333333	2.666667	-1.666667	2.0
emily	-3.5	0.000000	0.500000	1.500000	-3.5

states = ["ohio","new york","vermont","florida",
         "oregon","nevada","california","idaho"]
group_key = ["east"]*4 + ["west"]*4
data = pd.Series(np.random.randint(8),index = states)
data[["vermont","nevada","idaho"]] = np.nan
data

ohio          7.0
new york      7.0
vermont       NaN
florida       7.0
oregon        7.0
nevada        NaN
california    7.0
idaho         NaN
dtype: float64

data.groupby(group_key).mean()

east    7.0
west    7.0
dtype: float64

data.groupby(group_key).apply(lambda g: g.fillna(g.mean()))

ohio          7.0
new york      7.0
vermont       7.0
florida       7.0
oregon        7.0
nevada        7.0
california    7.0
idaho         7.0
dtype: float64