如何将数据框行转换为标题？

Question

I have some data on each of the first 151 pokemon in 151 different dataframes.

    id  identifier  pokemon_id  stat_id base_stat   local_language_id   name
36  7   Squirtle    7   1   44  9   HP
37  7   Squirtle    7   2   48  9   Attack
38  7   Squirtle    7   3   65  9   Defense
39  7   Squirtle    7   4   50  9   Special Attack
40  7   Squirtle    7   5   64  9   Special Defense
41  7   Squirtle    7   6   43  9   Speed


    id  identifier  pokemon_id  stat_id base_stat   local_language_id   name
18  4   Charmander  4   1   39  9   HP
19  4   Charmander  4   2   52  9   Attack
20  4   Charmander  4   3   43  9   Defense
21  4   Charmander  4   4   60  9   Special Attack
22  4   Charmander  4   5   50  9   Special Defense
23  4   Charmander  4   6   65  9   Speed

What I would really like is one row per pokemon with each stat as a column of a new dataframe. Something like

id    identifier    pokemon_id   HP  Attack    ...
4     Charmander    4            39  52        ...
7     Squirtle      7            44  48        ...

Is there an easy way to do that with a pandas dataframe?

Answer 1

我相信这会满足您的要求：

df.groupby(['id', 'identifier', 'name']).base_stat.first().unstack('name')

Answer 2

You can use pivot_table :

df = df.pivot_table(index=['id','identifier'], 
                    columns='name', 
                    values='base_stat', 
                    aggfunc='first')

print (df)
name           Attack  Defense  HP  Special Attack  Special Defense  Speed
id identifier                                                             
7  Squirtle        48       65  44              50               64     43

If all DataFrames are in list dfs , use concat with list comprehension :

dfs = [df1, df2]

df = pd.concat([df.pivot_table(index=['id','identifier'], 
                               columns='name', 
                               values='base_stat', 
                               aggfunc='first') for df in dfs])
print (df)
name           Attack  Defense  HP  Special Attack  Special Defense  Speed
id identifier                                                             
7  Squirtle        48       65  44              50               64     43
4  Charmander      52       43  39              60               50     65

Last use reset_index with rename_axis (new in pandas 0.18.0 ), if use pandas bellow 0.18.0 omit rename_axis and use df.columns.name = None :

df = pd.concat([df.pivot_table(index=['id','identifier'], 
                               columns='name', 
                               values='base_stat', 
                               aggfunc='first') for df in dfs])
                  .reset_index()
                  .rename_axis(None, axis=1)
print (df)
   id  identifier  Attack  Defense  HP  Special Attack  Special Defense  Speed
0   7    Squirtle      48       65  44              50               64     43
1   4  Charmander      52       43  39              60               50     65