Pandas 对数据和计数进行分组，包括值 0

Question

When I group and count the data, it will only show the number of existing values, which can be displayed intuitively, but I can't use it. My data:

name    group
aa  a
aa  b
bb  a
cc  b
dd  a
dd  b

My code：

df=pd.read_csv(fp6)
a=df.groupby(['name','group'])['group'].size()

result:

name  group
aa    a        1
      b        1
bb    a        1
cc    b        1
dd    a        1
      b        1

If I extract its value to draw a chart, it will prompt me that I have missing parameters. I want to show all 'group' values for each 'name'. like this:

name  group
aa    a        1
      b        1
bb    a        1
      b        0
cc    a        0
      b        1
dd    a        1
      b        1

Can someone teach me?

Answer 1

Use pd.crosstab :

>>> pd.crosstab(df['name'], df['group']).stack('group')
name  group
aa    a        1
      b        1
bb    a        1
      b        0
cc    a        0
      b        1
dd    a        1
      b        1
dtype: int64

Answer 2

你可以使用这个：

df = df.groupby(['name','group']).count().unstack(fill_value=0).stack()

Answer 3

A really explicit way to ensure having all combinations is to use Categorical data (conversion can by made using pandas.Categorical ).

NB. I added here a group 'c' in the categories that is non represented in the data to illustrate this point.

df['name'] = pd.Categorical(df['name'], categories=['aa', 'bb', 'cc', 'dd'])
df['group'] = pd.Categorical(df['group'], categories=['a', 'b', 'c'])

df.groupby(['name','group'])['group'].size()

output:

name  group
aa    a        1
      b        1
      c        0
bb    a        1
      b        0
      c        0
cc    a        0
      b        1
      c        0
dd    a        1
      b        1
      c        0
Name: group, dtype: int64