Python 将操作应用于所有字典键

给定数据帧

df

：

           paper    reference   count
9384155    p25      r50         1
7434371    p98      r9          78
7433400    p7       r27         5
7431765    p101     r91         501
7422256    p22      r5          91
...

我创建了一个字典，通过

count

将

df

子集：

df_dict={key:df[df['count']==key] for key in df['count'].unique()}

对于

df_dict

中的每个子数据帧，我希望应用以下操作：

df_dict[i] = df_dict[i].drop(['count'], axis=1)

pairs = df_dict[i].merge(df_dict[i], on=["reference"])
pairs = pairs[pairs["paper_x"] < pairs["paper_y"]]
pairs = pairs.groupby(["paper_x", "paper_y"]).count().reset_index()
pairs.columns = ["paper1", "paper2", "common"]

refs = df_dict[i].groupby(["paper"]).count().reset_index()
refs.columns = ["paper", "freq"]

result = pairs.merge(refs, how="left", left_on="paper1", right_on="paper")
result = result.merge(refs, how="left", left_on="paper2", right_on="paper")
result = result[["paper1", "freq_x", "paper2", "freq_y", "common"]]
result.columns = ["paper1", "freq1", "paper2", "freq2", "common"]

但是该操作返回一个大数据帧，并且返回的值不正确

---

理想情况下，我想要一组多个

result

数据帧（用

count

分隔，就像它们在

df_dict

中一样）。

“我尝试在范围内（df['count'].max（））为I运行一个循环。”请将此作为代码的一部分显示。抱歉，我认为这是非常冗余和冗长的。应用了编辑。“我尝试在范围内（df['count'].max（））为I运行循环…”请将此作为代码的一部分显示。抱歉，我认为这是非常冗余和冗长的。应用编辑。

for i in range(df['count'].max()):
  try:
    c = df_dict[i]
    c = c.drop(['count'], axis=1)
    pairs = c.merge(c,on=['reference'])
    pairs = pairs[pairs["paper_x"] < pairs["paper_y"]]
    pairs = pairs.groupby(["paper_x", "paper_y"]).count().reset_index() 
    pairs.columns = ["paper1", "paper2", "common"]    
    refs = c.groupby(["paper"]).count().reset_index()
    refs.columns = ["paper", "freq"]
    result = pairs.merge(refs, how="left", left_on="paper1", right_on="paper")
    result = result.merge(refs, how="left", left_on="paper2", right_on="paper")
    result = result[["paper1", "freq_x", "paper2", "freq_y", "common"]]
    result.columns = ["paper1", "freq1", "paper2", "freq2", "common"]
  except KeyError:
    continue