Question

这里我有一个包含交易的数据集。每个交易可以具有1个以上的不同值-“维度”。每个交易的值不能相同。我想创建一个在列和行中都带有“维度”的数据框，并计算一次交易中一个维度与另一个维度一起使用的次数。

这是我尝试过的

dim_set = [ (1, 'Customer group$Large'),
        (1, 'DEPARTMENT$Sales'),
        (2, 'Customer group$Medium'),
        (2, 'DEPARTMENT$Sales'),
        (3, 'DEPARTMENT$Sales'),
        (4, 'Customer group$Small'),
        (4, 'DEPARTMENT$Sales')
        ]
df = pd.DataFrame(dim_set, columns=['combination_id', 'dimension'])
df

df_st_1 = df.pivot_table(index = 'dimension', columns = 'dimension',values = 'combination_id', aggfunc = 'count')
df_st_1

预期结果应该是这样

dim_set = [ ('Customer group$Large', 1, 1, 0, 0),
        ('DEPARTMENT$Sales', 1, 4, 1, 1),
        ('Customer group$Medium', 0, 1, 1, 0),
        ('Customer group$Small', 0, 1, 0, 1)
        ]

df = pd.DataFrame(dim_set, columns=['dimension','Customer group$Large', 'DEPARTMENT$Sales', 'Customer group$Medium', 'Customer group$Small'])
df

Answer 1

将DataFrame.merge与crosstab一起使用，最后通过DataFrame.reset_index和DataFrame.rename_axis清除一些数据：

df1 = df.merge(df, on='combination_id', suffixes=('','_'))
df1 = (pd.crosstab(df1['dimension'], df1['dimension_'])
         .reset_index()
         .rename_axis(None)
         .rename_axis(None, axis=1))
print (df1)
               dimension  Customer group$Large  Customer group$Medium  \
0   Customer group$Large                     1                      0   
1  Customer group$Medium                     0                      1   
2   Customer group$Small                     0                      0   
3       DEPARTMENT$Sales                     1                      1   

   Customer group$Small  DEPARTMENT$Sales  
0                     0                 1  
1                     0                 1  
2                     1                 1  
3                     1                 4

如何在熊猫数据框中获取组合的组合？

1 个答案: