python - 连接多索引

Question

在区域和日期上合并 2 个数据框的最简单方法是什么？

我尝试加入并合并和连接。我收到“float”和“str”实例之间不支持“'<”和“无法处理非唯一多索引”错误

old_df                                  old_value
      region            date      
        England       2010-01-01         4
                      2010-01-02         5
         Wales        2010-01-01         4
                      2010-01-02         3
...
new_df 

                        name         new_value
      region            date      
        England       2010-01-01         10
                      2010-01-02         10
         Wales        2010-01-01         9
                      2010-01-02         10
...

预期产出

                                 old_value   new_value 
       region          date      
        England       2010-01-01         4         10
                      2010-01-02         5         10
         Wales        2010-01-01         4         9
                      2010-01-02         3         10

标签： pythonpython-3.xpandasmulti-index

Answer 1

完美运行。您确定您已确保您的日期列是日期吗？pd.to_datetime()

df_old = pd.read_csv(io.StringIO("""
      region            date      old_value
        England       2010-01-01         4
            nan          2010-01-02         5
         Wales        2010-01-01         4
            nan          2010-01-02         3
"""), sep="\s+")

df_new = pd.read_csv(io.StringIO("""        
      region            date       new_value
        England       2010-01-01         10
            nan          2010-01-02         10
         Wales        2010-01-01         9
             nan         2010-01-02         10"""), sep="\s+")

df_old.region = df_old.region.fillna(method="ffill")
df_new.region = df_new.region.fillna(method="ffill")
df_old.date = pd.to_datetime(df_old.date)
df_new.date = pd.to_datetime(df_new.date)

dfj = df_old.set_index(["region","date"]).join(df_new.set_index(["region","date"]))

                    old_value  new_value
region  date                            
England 2010-01-01          4         10
        2010-01-02          5         10
Wales   2010-01-01          4          9
        2010-01-02          3         10