获取百分比和计数 Python

Question

假设df.bun（df是一个Pandas数据帧）是一个多索引（日期和名称），变量是用字符串编写的类别值，

date      name             values
20170331  A122630          stock-a
          A123320          stock-a
          A152500          stock-b
          A167860          bond
          A196030          stock-a
          A196220          stock-a
          A204420          stock-a
          A204450          curncy-US
          A204480          raw-material
          A219900          stock-a

我怎样才能使它代表同一日期的总计数及其百分比，使 table 像下面的每个日期一样，

date           variable    counts     Percentage
20170331          stock         7           70%
                   bond         1           10%
           raw-material         1           10%
                 curncy         1           10%

我已经print(df.groupby('bun').count())解决了这个问题，但它缺少..

cf) 在获取 df.bun 之前，我使用以下代码将嵌套字典导入 Pandas 数据框。

import numpy as np
import pandas as pd

result = pd.DataFrame()
origDict = np.load("Hannah Lee.npy")
for item in range(len(origDict)):
    newdict = {(k1, k2):v2 for k1,v1 in origDict[item].items() for k2,v2 in origDict[item][k1].items()}
    df = pd.DataFrame([newdict[i] for i in sorted(newdict)],
                      index=pd.MultiIndex.from_tuples([i for i in sorted(newdict.keys())]))
    print(df.bun)

Answer 1

我认为需要 SeriesGroupBy.value_counts:

g = df.groupby('date')['values']
df = pd.concat([g.value_counts(), 
                g.value_counts(normalize=True).mul(100)],axis=1, keys=('counts','percentage'))
print (df)
                       counts  percentage
date     values                          
20170331 stock-a            6        60.0
         bond               1        10.0
         curncy-US          1        10.0
         raw-material       1        10.0
         stock-b            1        10.0

size for counts and then divide by new Series created by transform 和 sum 的另一个解决方案：

df2 = df.reset_index().groupby(['date', 'values']).size().to_frame('count')
df2['percentage'] = df2['count'].div(df2.groupby('date')['count'].transform('sum')).mul(100)
print (df2)
                       count  percentage
date     values                         
20170331 bond              1        10.0
         curncy-US         1        10.0
         raw-material      1        10.0
         stock-a           6        60.0
         stock-b           1        10.0

解决方案之间的区别是首先按每组的值排序，然后再排序 MultiIndex。

获取百分比和计数 Python

getting percentage and count Python

python

percentage

pandas

pandas-groupby