python - 如何使用 pandas groupby 对某些行进行降序排序，对某些行进行升序排序

Question

我有一个像这样的 DataFrame:

import pandas as pd
df = pd.DataFrame(
    [
              ["EUR"          , 0.031        , 61.170000 , "ask" ],
              ["EUR"          , 7.642        , 61.360000 , "ask" ],
              ["EUR"          , 0.266        , 61.370000 , "ask" ],
              ["EUR"          , 0.364        , 61.410000 , "ask" ],
              ["EUR"          , 0.101        , 61.100000 , "bid" ],
              ["EUR"          , 23.090       , 60.470000 , "bid" ],
              ["EUR"          , 2.061        , 60.460000 , "bid" ],
              ["EUR"          , 0.133        , 60.450000 , "bid" ],
              ["USD"          , 3.031        , 161.170000, "ask" ],
              ["USD"          , 10.642       , 161.360000, "ask" ],
              ["USD"          , 3.266        , 161.370000, "ask" ],
              ["USD"          , 3.364        , 161.410000, "ask" ],
              ["USD"          , 3.101        , 161.100000, "bid" ],
              ["USD"          , 26.090       , 160.470000, "bid" ],
              ["USD"          , 5.061        , 160.460000, "bid" ],
              ["USD"          , 3.133        , 160.450000, "bid" ],
    ],
    columns = ["base_currency", "base_volume", "price"  , "type"]
)
df

我想重新排列行。具体来说，我想对货币进行分组(所有“EUR”行放在一起，所有“USD”行放在一起)然后，对于每种货币，我希望“询问”按降序排列，我希望“出价”是升序排列。现在，如果 DataFrame 仅包含一种货币，我可以通过以下方式正常工作:

_df = df.groupby("type").apply(lambda row: row.sort_values(by = "price"))
_df

我必须处理多种货币，但是当我尝试按以下方式对它们进行分组时

_df = df.groupby(["base_currency", "type"]).apply(
    lambda row:
        row.sort_values(by = "price") if row['type'] == 'ask' else\
        row.sort_values(by = "price", ascending = False),
    axis = 1
)
_df

我收到以下错误:

TypeError: <lambda>() got an unexpected keyword argument 'axis'

这里可能发生了什么？我该如何解决？

Answer 1

发生的事情是 groupby.apply 没有错误提示的 axis 参数，您可以改为这样做:

df.groupby(
    ['base_currency', 'type'], group_keys=False
).apply(
    lambda g: g.sort_values('price', ascending=g.name[1] == 'bid')     
    # check the type for each group and sort accordingly
)

#   base_currency  base_volume   price type
#3            EUR        0.364   61.41  ask
#2            EUR        0.266   61.37  ask
#1            EUR        7.642   61.36  ask
#0            EUR        0.031   61.17  ask
#7            EUR        0.133   60.45  bid
#6            EUR        2.061   60.46  bid
#5            EUR       23.090   60.47  bid
#4            EUR        0.101   61.10  bid
#11           USD        3.364  161.41  ask
#10           USD        3.266  161.37  ask
#9            USD       10.642  161.36  ask
#8            USD        3.031  161.17  ask
#15           USD        3.133  160.45  bid
#14           USD        5.061  160.46  bid
#13           USD       26.090  160.47  bid
#12           USD        3.101  161.10  bid