使用Pandas將列添加到數據透視表

Question

我的桌子如下：

import pandas as pd
import numpy as np

#simple table
fazenda = [6010,6010,6010,6010]
quadra = [1,1,2,2]
talhao = [1,2,3,4]
arTotal = [32.12,33.13,34.14,35.15]
arCarr = [i/2 for i in arTotal]
arProd = [i/2 for i in arTotal]
varCan = ['RB1','RB2','RB3','RB4']
data = list(zip(fazenda,quadra,talhao,arTotal,arCarr,arProd,varCan))

#Pandas DataFrame
df = pd.DataFrame(data=data,columns=['Fazenda','Quadra','Talhao','ArTotal','ArCarr','ArProd','Variedade'])

#Pivot Table
table = pd.pivot_table(df, values=['ArTotal','ArCarr','ArProd'],index=['Quadra','Talhao'], fill_value=0)

print(table)

結果是：

               ArCarr  ArProd  ArTotal
Quadra Talhao                         
1      1       16.060  16.060    32.12
       2       16.565  16.565    33.13
2      3       17.070  17.070    34.14
       4       17.575  17.575    35.15

我需要兩個附加步驟：

1. 為“ ArTotal”，“ ArCarr”和“ ArProd”字段添加小計和總計
2. 在表格中添加“變量”字段

我試圖添加列，但結果不正確。 跟隨有關Total和Grand Total的一些鏈接，我沒有得到令人滿意的結果。

我很難理解大熊貓，我要求經驗豐富的同事幫忙。

Answer 1

首先獲得正確的pivot 。

In [404]: values = ['ArTotal','ArCarr','ArProd']

In [405]: table = pd.pivot_table(df, values=values, index=['Quadra','Talhao','Variedade'], 
                                 fill_value=0).reset_index(level=-1)

獲取總計

In [406]: Gt = table[values].sum()

獲取Quadra級總計

In [407]: St = table.sum(level='Quadra')

使用append重塑table

In [408]: (table.append(
                 St.assign(Talhao='Total').set_index('Talhao', append=True)
                ).sort_index()
                .append(pd.DataFrame([Gt.values], columns=Gt.index,
                                     index=pd.MultiIndex.from_tuples([('Grand Total', '')],
                                     names=['Quadra', 'Talhao']))
                ).fillna(''))
Out[408]:
                    ArCarr  ArProd  ArTotal Variedade
Quadra      Talhao
1           1       16.060  16.060    32.12       RB1
            2       16.565  16.565    33.13       RB2
            Total   32.625  32.625    65.25
2           3       17.070  17.070    34.14       RB3
            4       17.575  17.575    35.15       RB4
            Total   34.645  34.645    69.29
Grand Total         67.270  67.270   134.54

細節

In [409]: table
Out[409]:
              Variedade  ArCarr  ArProd  ArTotal
Quadra Talhao
1      1            RB1  16.060  16.060    32.12
       2            RB2  16.565  16.565    33.13
2      3            RB3  17.070  17.070    34.14
       4            RB4  17.575  17.575    35.15

In [410]: Gt
Out[410]:
ArTotal    134.54
ArCarr      67.27
ArProd      67.27
dtype: float64

In [411]: St
Out[411]:
        ArCarr  ArProd  ArTotal
Quadra
1       32.625  32.625    65.25
2       34.645  34.645    69.29

Answer 2

我認為John的解決方案擊敗了我，但是根據您當前的輸出，您不能使用數據透視表來做到這一點，您可以使用分組數據的列表理解來執行一系列步驟，然后附加總和來做到這一點，即

cols = ['Fazenda','Variedade','Quadra','Talhao']
ndf = pd.concat([i.append(i.drop(cols,1).sum(),1) for _,i in df.groupby('Quadra')])

ndf['Talhao'] = ndf[['Talhao']].fillna('Total')
ndf['Quadra'] = ndf['Quadra'].ffill()

new = ndf.set_index(['Quadra','Talhao']).drop(['Fazenda'],1)

new = new.append(pd.DataFrame(df.sum()).T.drop(cols,1).set_index(pd.MultiIndex.from_tuples([('Grand Total', '')]))).fillna('')

輸出：

ArCarr  ArProd  ArTotal Variedade
Quadra      Talhao                                   
1.0         1.0     16.060  16.060    32.12       RB1
            2.0     16.565  16.565    33.13       RB2
            Total   32.625  32.625    65.25          
2.0         3.0     17.070  17.070    34.14       RB3
            4.0     17.575  17.575    35.15       RB4
            Total   34.645  34.645    69.29          
Grand Total         67.270  67.270   134.54