首页 文章

熊猫以相等的部分重新采样时间系列

提问于
浏览 次
1

我试图在 N 等份中重新取样一只大熊猫时间系列 .

我的时间系列大小 10 ：

rng = pd.date_range('20130101',periods=10,freq='T')
ts=pd.Series(np.random.randn(len(rng)), index=rng)

print(ts)
2013-01-01 00:00:00   -1.346024
2013-01-01 00:01:00    0.671637
2013-01-01 00:02:00    0.435566
2013-01-01 00:03:00    1.043379
2013-01-01 00:04:00    0.877782
2013-01-01 00:05:00   -1.216598
2013-01-01 00:06:00    0.801525
2013-01-01 00:07:00    1.041827
2013-01-01 00:08:00   -0.309048
2013-01-01 00:09:00    0.212750
Freq: T, dtype: float64

How can I resample in N equal parts, N being > 10

我希望答案当然有一些NaN .

我希望看起来像（N = 12）：

2013-01-01 00:00:00   -1.346024
2013-01-01 00:00:50    0.671637
2013-01-01 00:01:40    0.435566
2013-01-01 00:02:30    1.043379
2013-01-01 00:03:20    0.877782
2013-01-01 00:04:10         NaN
2013-01-01 00:05:00   -1.216598
2013-01-01 00:05:50    0.801525
2013-01-01 00:06:40    1.041827
2013-01-01 00:07:30   -0.309048
2013-01-01 00:08:20    0.212750
2013-01-01 00:09:10         NaN
Freq: 50S, dtype: float64

注意：以下代码适用于N <10：

length = 9
timeSpan = (ts.index[-1]-ts.index[0]+timedelta(minutes=1))
rule = int(timeSpan.total_seconds()/length)
tsNew=ts.resample(str(rule)+"S")

print(tsNew)
2013-01-01 00:00:00   -0.337194
2013-01-01 00:01:06    0.435566
2013-01-01 00:02:12    1.043379
2013-01-01 00:03:18    0.877782
2013-01-01 00:04:24   -1.216598
2013-01-01 00:05:30    0.801525
2013-01-01 00:06:36    1.041827
2013-01-01 00:07:42   -0.309048
2013-01-01 00:08:48    0.212750
Freq: 66S, dtype: float64

注意：如果可能，我的答案仍然与多列时间序列兼容（例如开放/高/低/关闭金融系列）

python pandas time-series financial resampling

1 回答

  • 0

    解决了 . 重新采样方法应使用fill_method ='pad'和closed ='right'

    tsNew=ts.resample(str(rule)+"S", fill_method='pad',closed='right')

ts
2013-01-01 00:00:00   -1.827784
2013-01-01 00:01:00   -2.181001
2013-01-01 00:02:00   -2.498234
2013-01-01 00:03:00   -0.646579
2013-01-01 00:04:00   -0.720016
2013-01-01 00:05:00    1.298624
2013-01-01 00:06:00   -0.785790
2013-01-01 00:07:00    0.769829
2013-01-01 00:08:00   -0.877086
2013-01-01 00:09:00   -0.311500
Freq: T, dtype: float64

tsNew (for N=12)
2013-01-01 00:00:00   -1.827784
2013-01-01 00:00:50   -1.827784
2013-01-01 00:01:40   -2.181001
2013-01-01 00:02:30   -2.498234
2013-01-01 00:03:20   -0.646579
2013-01-01 00:04:10   -0.720016
2013-01-01 00:05:00    1.298624
2013-01-01 00:05:50    1.298624
2013-01-01 00:06:40   -0.785790
2013-01-01 00:07:30    0.769829
2013-01-01 00:08:20   -0.877086
2013-01-01 00:09:10   -0.311500
Freq: 50S, dtype: float64
    回复于

相关问题