[英]Use pandas groupby function on multiple columns
I have a DataFrame similar to this: 我有一个类似于以下的DataFrame:
Key Departure Species1 Species2 Status
1 R Carlan Carlan D
1 R Scival Carex C
2 R Carlan Scival D
2 R Scival Bougra C
3 D Carlan Carlan D
3 D Scival Scival C
I want to count the occurrences of each unique Species1
for a given Departure
and Status
of D
of C
我想计算
C
的D
的给定Departure
和Status
下每个唯一Species1
的出现
My desired output is: 我想要的输出是:
Species1 RD RC DD DC
Carlan 2 NaN 1 NaN
Scival NaN 2 NaN 1
Make a new column that is the combination of Departure and Status 新建一个包含“出发时间”和“状态”的组合的列
df['comb'] = df.Departure + df.Status
df
# Key Departure Species1 Species2 Status comb
#0 1 R Carlan Carlan D RD
#1 1 R Scival Carex C RC
#2 2 R Carlan Scival D RD
#3 2 R Scival Bougra C RC
#4 3 D Carlan Carlan D DD
#5 3 D Scival Scival C DC
Then you can groupby: 然后,您可以分组:
gb = df.groupby(['Species1', 'comb'])
gb.groups
#{('Carlan', 'DD'): [4],
#('Carlan', 'RD'): [0, 2],
#('Scival', 'DC'): [5],
#('Scival', 'RC'): [1, 3]}
Now organize the results into a list, where each element is a tuple (column, Series(data, index))
representing a single data point in a new dataframe 现在将结果组织成一个列表,其中每个元素都是一个元组
(column, Series(data, index))
表示新数据帧中的单个数据点
items = [ (key[1], pandas.Series( [len(val)], index=[key[0]] ) )for key,val in gb.groups.items() ]
And make a new dataframe from the items: 并从以下各项创建一个新的数据框:
result = pandas.from_items( items)
result
# RC DC DD RD
#Carlan NaN NaN 1 2
#Scival 2 1 NaN NaN
See this link for ideas on crating new dataframes from various objects. 请参阅此链接,以获取有关从各种对象创建新数据框的想法。 When you want to create a dataframe from individual data points (eg (Species1,comb) ), then
from_items
is your best option. 当您要根据单个数据点(例如(Species1,comb))创建数据框时,
from_items
是最佳选择。
Use the pandas.crosstab() method. 使用pandas.crosstab()方法。 A single line of code:
一行代码:
pd.crosstab(df.Species1, [df.Departure, df.Status])
The resulting table: 结果表:
If you combine with @dermen's 'comb' column, 如果与@dermen的“梳子”列结合使用,
df['comb'] = df.Departure + df.Status
pd.crosstab(df.Species1, df.comb)
you'll get: 你会得到:
If you really want those 'NaN', just tack on a .replace('0', np.nan)
, like so (assuming an import numpy as np
has already been done): 如果您真的想要那些'NaN',只需在
.replace('0', np.nan)
,就像这样(假设已经完成了import numpy as np
已经完成):
pd.crosstab(df.Species1, df.comb).replace('0', np.nan)
您可以对多个列使用groupby查询,并使用.agg函数来计算出现次数:
df.groupby(['Species1', 'Departure', 'Status']).agg(['count'])
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.