在Pandas中将通用float64索引转换为Timestamp

Question

Consider the following series, how can I convert the index of the following series to a Timestamp, assuming that the index is in seconds ?

quantity
1631.083        -5000
1632.395        -5000
1635.482        -5000
1638.536        -1800
1640.818        -5000
1644.739        -5000
1644.828        -5000
1655.214        -1800
1658.691        -4300
1662.751        -5000
1693.350        -5000
Length: 87575, dtype: float64

I tried:

pd.Timestamp(my_series.index)

but I get:

ValueError: Cannot convert Period to Timestamp unambiguously. Use to_timestamp

The final goal is to be able to use pd.resample() to resample from the series above.

Answer 1

Using today's date as the base, you can convert the floats to Timedelta values, interpreting the float values as seconds, and then adding that to today's date:

In [15]: s2 = s.copy()

In [16]: s2.index = pd.Timestamp(datetime.date.today()) + pd.TimedeltaIndex(s.index, unit='s')

In [17]: s2
Out[17]: 
2014-12-30 00:27:11.083000   -5000
2014-12-30 00:27:12.395000   -5000
2014-12-30 00:27:15.482000   -5000
2014-12-30 00:27:18.536000   -1800
2014-12-30 00:27:20.818000   -5000
2014-12-30 00:27:24.739000   -5000
2014-12-30 00:27:24.828000   -5000
2014-12-30 00:27:35.214000   -1800
2014-12-30 00:27:38.691000   -4300
2014-12-30 00:27:42.751000   -5000
2014-12-30 00:28:13.350000   -5000
Name: quantity, dtype: int64