pandas 根据前缀重塑 long

Question

我有一个包含以下列的 Pandas 数据框

game_id, date, country, winner_name, winner_age, ... winner_ranking, loser_name, loser_age, ... loser_ranking
1        1/2/10  UK .     Ben          21               12            Michael     22 .    13

我想将其重塑为以下格式

game_id, date, country, competitor, name, age, ranking 
 1       1/2/10 UK       winner    Ben    21   12
 1       1/2/10 UK       loser     Michael 22   13

即对于以前缀 'winner_' 或 'loser_' 开头的每一列，删除此前缀，并将赢家和输家分成不同的行。赢家和输家变量的列表很长，所以如果我必须硬编码，它就没有多大帮助了。

这是我目前的做法，我想知道是否有更简洁的方法，例如使用 melt？

winner_df = combined_df.loc[:,[x for x in colnames if 'loser_' not in x]]
winner_df.columns = [c.replace('winner_','') for c in winner_df.columns]
winner_df['competitor'] = 'winner'
loser_df = combined_df.loc[:,[x for x in colnames if 'winner_' not in x]]
loser_df.columns = [c.replace('loser_','') for c in loser_df.columns]
loser_df['competitor'] = 'loser'
long_df = winner_df.append(loser_df,sort=False)

Answer 1

首先从所有没有带拆分器的列的列创建 MultiIndex DataFrame.set_index, then create MultiIndex in columns by Series.str.split and last reshape by DataFrame.stack with DataFrame.reset_index 和 rename 列：

df = df.set_index(['game_id','date','country'])

df.columns = df.columns.str.split('_', expand=True)
df = df.stack(0).reset_index().rename(columns={'level_3':'competitor'})
print (df) 
   game_id    date country competitor  age     name  ranking
0        1  1/2/10      UK      loser   22  Michael       13
1        1  1/2/10      UK     winner   21      Ben       12

pandas 根据前缀重塑 long

pandas reshape long based on prefixes

python

reshape

pandas

data-cleaning