Pandas-数据透视表由2列生成

Question

I am trying to generate a pivot table structure but only from two columns of data. What I have is this general DataFrame;

df = pd.DataFrame({'name': ['Australia', 'Japan', 'Brazil'], 'code': ['R1', 'R2', 'R3']})

and what I am trying to achieve is have the name field in both the columns and the index, and the values to the the concatenated strings from the code field. This will result in a DataFrame with shape (3, 3). Essentially having in the Australia row and Japan column the value of R1-R2 and in the Brazil row and Australia column the value of R3-R1 .

I tried using this function but I am not sure the pivot_table can take the same value in the index and columns.

pd.pivot_table(df, values='code', index=['name'], columns=['name'], aggfunc=lambda x: '-'.join(x))

Essentially, the output should be of this form (although maybe with index and column names) and not manually generated;

data = {'Australia': ['R1-R1', 'R2-R1', 'R3-R1'],
        'Japan': ['R1-R2', 'R2-R2', 'R3-R2'],
        'Brazil': ['R1-R3', 'R2-R3', 'R3-R3']}

df_result = pd.DataFrame(data, columns=['Australia', 'Japan', 'Brazil'], index=['Australia', 'Japan', 'Brazil'])

Thanks!

Answer 1

One way, you could do this:

df1 = df.assign(key=1).merge(df.assign(key=1), how='outer',on='key',suffixes=('','_c'))

df1 = df1.drop('key',axis=1)

df1['value'] = df1['code'] + '-' + df1['code_c']

df2 = df1.drop(['code','code_c'],axis=1)

df_result = df2.set_index(['name','name_c']).unstack()

df_result.columns = df_result.columns.droplevel()

print(df_result)

Output:

name_c    Australia Brazil  Japan
name                             
Australia     R1-R1  R1-R3  R1-R2
Brazil        R3-R1  R3-R3  R3-R2
Japan         R2-R1  R2-R3  R2-R2