Python Pandas : float values break down when df. 列 = df.columns.droplevel()

Question

我有一个包含交易的数据框。我想从这个 OHLCV 中获得。

当我为此聚合时，我得到一个多索引数据帧。当我通过列的分配从多索引降低级别时，浮点尾部会出现额外的字符。

import pandas as pd

trades = pd.DataFrame(
    data={
        'price': np.random.uniform(1000, 1200, 48),
        'volume': np.random.uniform(50, 100, 48)
    },
    index=pd.date_range(start='2019-01-01 00:00', end='2019-01-02 23:00', freq='H')
)

ohlcv = trades.resample('D').agg({'price': 'ohlc', 'volume': 'sum'})
# ohlcv['volume'].iloc[0] == 1744.871219

ohlcv.columns = ohlcv.columns.droplevel()
# ohlcv['volume'].iloc[0] == 1744.8712193218892

• 如何去掉多级索引以便 float 数字不变？或
• 如何聚合初始数据帧而不生成多索引？

Answer 1

没有改变值，只是没有显示，因为第一个iloc返回一个值Series:

np.random.seed(456)
trades = pd.DataFrame(
    data={
        'price': np.random.uniform(1000, 1200, 48),
        'volume': np.random.uniform(50, 100, 48)
    },
    index=pd.date_range(start='2019-01-01 00:00', end='2019-01-02 23:00', freq='H') )

ohlcv = trades.resample('D').agg({'price': 'ohlc', 'volume': 'sum'})
print (ohlcv)
                  price                                              volume
                   open         high          low        close       volume
2019-01-01  1049.751182  1177.140381  1001.636683  1001.636683  1777.630832
2019-01-02  1072.013700  1194.631131  1008.525926  1130.678109  1804.651706

print (ohlcv['volume'].iloc[0])
volume    1777.630832
Name: 2019-01-01 00:00:00, dtype: float64

print (ohlcv['volume'].iloc[0].tolist())
[1777.630832239808]

ohlcv.columns = ohlcv.columns.droplevel()
print (ohlcv['volume'].iloc[0])
1777.630832239808

删除 MultiIndex 的另一个解决方案是 map 和 join:

ohlcv.columns = ohlcv.columns.map('_'.join)
print (ohlcv)
             price_open   price_high    price_low  price_close  volume_volume
2019-01-01  1049.751182  1177.140381  1001.636683  1001.636683    1777.630832
2019-01-02  1072.013700  1194.631131  1008.525926  1130.678109    1804.651706

使用 MultiIndex.droplevel 的解决方案也可以使用，但要小心避免返回重复的列名:

ohlcv = trades.resample('D').agg({'price': 'ohlc', 'volume': 'ohlc'})
print (ohlcv)
                  price                                            volume  \
                   open         high          low        close       open   
2019-01-01  1049.751182  1177.140381  1001.636683  1001.636683  68.335978   
2019-01-02  1072.013700  1194.631131  1008.525926  1130.678109  85.631147   


                 high        low      close  
2019-01-01  98.772905  50.545352  70.416061  
2019-01-02  99.011524  51.194741  57.057429  

ohlcv.columns = ohlcv.columns.droplevel()
print (ohlcv)
                   open         high          low        close       open  \
2019-01-01  1049.751182  1177.140381  1001.636683  1001.636683  68.335978   
2019-01-02  1072.013700  1194.631131  1008.525926  1130.678109  85.631147   

                 high        low      close  
2019-01-01  98.772905  50.545352  70.416061  
2019-01-02  99.011524  51.194741  57.057429 

print (ohlcv['open'])
                   open       open
2019-01-01  1049.751182  68.335978
2019-01-02  1072.013700  85.631147