Question

我有一堆刻度数据，我可以使用以下方法成功重新采样到时间数据中：

h5_file = pd.HDFStore(h5_path)
h5_file['fx_data'].groupby('Symbol')
ask = grouped['Ask'].resample('5Min', how='ohlc')
bid = grouped['Bid'].resample('5Min', how='ohlc')

但我想也要回复音量。这应该只是每个样本组成的行数。怎样才能做到最好？

此外 - 当我选择使用较小的时间帧重新采样时，偶尔会出现值为N / A的条形码，因为该期间没有价格变化。当发生这种情况时，我希望前一个接近是当前柱上OHLC的值。

我已经搜索过并找到了这段代码：

whatev.groupby('Symbol')closes = resampledData['close'].fillna(method='pad')
resampledData.apply(lambda x: x.fillna(closes)

我对Python和编程非常陌生，并且还不了解lambas。这是否只会改变我需要改变的接近值或所有值。非常感谢所有帮助。

Answer 1

我已经在hdf5中获得了部分样本外汇数据（2015年5月为美元/欧元）的副本，因此我将在此处用于说明目的。

import pandas as pd

Jian_h5 = '/media/Primary Disk/Jian_Python_Data_Storage.h5'
h5_file = pd.HDFStore(Jian_h5)  

fx_df = h5_file['fx_tick_data']
# I've only got USD/EUR in this dataset, but let's still do a groupby symbol
# and assume you have multiple symbols
grouped = fx_df.groupby('Symbol')

# calculate sub-group average bid and ask price, and also number of ticks
freq = '1min'
# an empty DataFrame
result = pd.DataFrame()
# bid/ask price: forward fill make sense
result['avg_bid'] = grouped['Bid'].resample(freq, how='mean').fillna(method='ffill')
result['avg_ask'] = grouped['Ask'].resample(freq, how='mean').fillna(method='ffill')
# tick count: NaN should be replaced by zero
result['tick_counts'] = grouped['Ask'].resample(freq, how='count').fillna(0)

Out[59]: 
                             avg_bid  avg_ask  tick_counts
Symbol  Date_time                                         
EUR/USD 2015-05-01 00:00:00   1.1210   1.1210           77
        2015-05-01 00:01:00   1.1209   1.1210          117
        2015-05-01 00:02:00   1.1209   1.1210           95
        2015-05-01 00:03:00   1.1210   1.1210           46
        2015-05-01 00:04:00   1.1211   1.1211          112
        2015-05-01 00:05:00   1.1213   1.1213          193
        2015-05-01 00:06:00   1.1214   1.1215           76
        2015-05-01 00:07:00   1.1216   1.1216          103
        2015-05-01 00:08:00   1.1216   1.1217          107
        2015-05-01 00:09:00   1.1217   1.1217           17
        2015-05-01 00:10:00   1.1216   1.1217           33
        2015-05-01 00:11:00   1.1218   1.1218           56
        2015-05-01 00:12:00   1.1217   1.1218           77
        2015-05-01 00:13:00   1.1215   1.1215           18
        2015-05-01 00:14:00   1.1215   1.1216           50
...                              ...      ...          ...
        2015-05-31 23:45:00   1.0959   1.0960           37
        2015-05-31 23:46:00   1.0959   1.0959           59
        2015-05-31 23:47:00   1.0958   1.0959           62
        2015-05-31 23:48:00   1.0956   1.0957           45
        2015-05-31 23:49:00   1.0955   1.0956           67
        2015-05-31 23:50:00   1.0955   1.0956           36
        2015-05-31 23:51:00   1.0955   1.0956           35
        2015-05-31 23:52:00   1.0956   1.0956           22
        2015-05-31 23:53:00   1.0956   1.0957           29
        2015-05-31 23:54:00   1.0957   1.0958           50
        2015-05-31 23:55:00   1.0956   1.0957           30
        2015-05-31 23:56:00   1.0957   1.0958            8
        2015-05-31 23:57:00   1.0957   1.0958           45
        2015-05-31 23:58:00   1.0957   1.0958           38
        2015-05-31 23:59:00   1.0958   1.0958           30

[44640 rows x 3 columns]

Python / Pandas重新采样外汇交易数据以获取交易量

1 个答案: