时间增量(Timedeltas)

The Timedeltas docs.

Using timedeltas

  1. In [175]: s  = pd.Series(pd.date_range('2012-1-1', periods=3, freq='D'))
  3. In [176]: s - s.max()
  4. Out[176]: 
  5. 0   -2 days
  6. 1   -1 days
  7. 2    0 days
  8. dtype: timedelta64[ns]
  10. In [177]: s.max() - s
  11. Out[177]: 
  12. 0   2 days
  13. 1   1 days
  14. 2   0 days
  15. dtype: timedelta64[ns]
  17. In [178]: s - datetime.datetime(2011,1,1,3,5)
  18. Out[178]: 
  19. 0   364 days 20:55:00
  20. 1   365 days 20:55:00
  21. 2   366 days 20:55:00
  22. dtype: timedelta64[ns]
  24. In [179]: s + datetime.timedelta(minutes=5)
  25. Out[179]: 
  26. 0   2012-01-01 00:05:00
  27. 1   2012-01-02 00:05:00
  28. 2   2012-01-03 00:05:00
  29. dtype: datetime64[ns]
  31. In [180]: datetime.datetime(2011,1,1,3,5) - s
  32. Out[180]: 
  33. 0   -365 days +03:05:00
  34. 1   -366 days +03:05:00
  35. 2   -367 days +03:05:00
  36. dtype: timedelta64[ns]
  38. In [181]: datetime.timedelta(minutes=5) + s
  39. Out[181]: 
  40. 0   2012-01-01 00:05:00
  41. 1   2012-01-02 00:05:00
  42. 2   2012-01-03 00:05:00
  43. dtype: datetime64[ns]

Adding and subtracting deltas and dates

  1. In [182]: deltas = pd.Series([ datetime.timedelta(days=i) for i in range(3) ])
  3. In [183]: df = pd.DataFrame(dict(A = s, B = deltas)); df
  4. Out[183]: 
  5.            A      B
  6. 0 2012-01-01 0 days
  7. 1 2012-01-02 1 days
  8. 2 2012-01-03 2 days
  10. In [184]: df['New Dates'] = df['A'] + df['B'];
  12. In [185]: df['Delta'] = df['A'] - df['New Dates']; df
  13. Out[185]: 
  14.            A      B  New Dates   Delta
  15. 0 2012-01-01 0 days 2012-01-01  0 days
  16. 1 2012-01-02 1 days 2012-01-03 -1 days
  17. 2 2012-01-03 2 days 2012-01-05 -2 days
  19. In [186]: df.dtypes
  20. Out[186]: 
  21. A             datetime64[ns]
  22. B            timedelta64[ns]
  23. New Dates     datetime64[ns]
  24. Delta        timedelta64[ns]
  25. dtype: object

Another example

Values can be set to NaT using np.nan, similar to datetime

  1. In [187]: y = s - s.shift(); y
  2. Out[187]: 
  3. 0      NaT
  4. 1   1 days
  5. 2   1 days
  6. dtype: timedelta64[ns]
  8. In [188]: y[1] = np.nan; y
  9. Out[188]: 
  10. 0      NaT
  11. 1      NaT
  12. 2   1 days
  13. dtype: timedelta64[ns]