熊貓唯一值作為帶計數的列

Question

使用 pandas 數據框並嘗試在分組輸出中翻轉它，該輸出采用唯一值並將它們作為一列，並將每個對應的計數作為新數據框中的值。

這是起始數據框：

df = pd.DataFrame([('gold', 'bronze', 'silver'),
                   ('silver', 'gold', 'bronze'),
                   ('gold', 'silver', 'bronze'),
                   ('bronze', 'silver', 'gold')],
                    columns=('Canada', 'China', 'South Korea'))
df.head()

    Canada  China   South Korea
0   gold    bronze  silver
1   silver  gold    bronze
2   gold    silver  bronze
3   bronze  silver  gold

所需的輸出是這樣的：

    nation      gold    silver  bronze
0   Canada        2          1       1
1   China         1          2       1
2   South Korea   1          1       2

Answer 1

您可以將df.apply與pd.value_counts df.apply使用^*

df.apply(pd.value_counts).T

             bronze  gold  silver
Canada            1     2       1
China             1     1       2
South Korea       2     1       1

---

^{* 我沒有找到pd.value_counts文檔，因此，將 github 鏈接鏈接到該函數。}

^{編輯：在閱讀源代碼pd.Series.value_counts只是調用pd.value_counts}

Answer 2

使用pd.get_dummies和sum

pd.get_dummies(df.T, prefix='',prefix_sep='').sum(level=0,axis=1)

Out[995]:
             bronze  gold  silver
Canada            1     2       1
China             1     1       2
South Korea       2     1       1

Answer 3

w = df.melt()

    variable    value
0   Canada      gold
1   Canada      silver
2   Canada      gold
3   Canada      bronze
4   China       bronze
5   China       gold
6   China       silver
7   China       silver
8   South Korea silver
9   South Korea bronze
10  South Korea bronze
11  South Korea gold

進而：

pd.crosstab(w['variable'],w['value'])

想要的結果：

value        bronze gold    silver
variable            
Canada        1      2       1
China         1      1       2
South Korea   2      1       1

Answer 4

df = pd.DataFrame([('gold', 'bronze', 'silver'),
               ('silver', 'gold', 'bronze'),
               ('gold', 'silver', 'bronze'),
               ('bronze', 'silver', 'gold')],
                columns=('Canada', 'China', 'South Korea')).transpose()

df.apply(pd.value_counts,axis=1)