如何在数据框中的行之间进行平均以填充缺失值[重复]答案

【问题标题】：How to average between rows in a Dataframe to fillup missing values [duplicate]如何在数据框中的行之间进行平均以填充缺失值[重复]
【发布时间】：2021-12-31 13:08:27
【问题描述】：

我有一个熊猫数据框：

Date
2016-11-21    26.292355
2016-11-22    26.308828
2016-11-23    26.174692
2016-11-24          NaN
2016-11-25    26.306471
2016-11-26          NaN
2016-11-27          NaN
2016-11-28    26.254705
2016-11-29    26.228815
2016-11-30    26.007618
2016-12-01    25.765236
2016-12-02    25.861721
2016-12-03          NaN
2016-12-04          NaN
2016-12-05    25.675812
2016-12-06    25.873482
2016-12-07    26.127634
2016-12-08    26.384132
2016-12-09    26.814764
Name: Close, dtype: object

我想用上一行和下一行的平均值来填充这些缺失的NaN 值。

生成的数据框将是：

Date
2016-11-21    26.292355
2016-11-22    26.308828
2016-11-23    26.174692
2016-11-24    26.240581  # Averaged value from previous and next row
2016-11-25    26.306471
2016-11-26    26.280588  # Averaged value from previous and next row
2016-11-27    26.280588  # Averaged value from previous and next row
2016-11-28    26.254705
2016-11-29    26.228815
2016-11-30    26.007618
2016-12-01    25.765236
2016-12-02    25.861721
2016-12-03    25.768766  # Averaged value from previous and next row
2016-12-04    25.768766  # Averaged value from previous and next row
2016-12-05    25.675812
2016-12-06    25.873482
2016-12-07    26.127634
2016-12-08    26.384132
2016-12-09    26.814764
Name: Close, dtype: object

我将如何在 python 中做到这一点？

【问题讨论】：

标签： python pandas dataframe interpolation nan

【解决方案1】：

使用Series.interpolate:

>>> sr.interpolate(method='linear')
Date
2016-11-21    26.292355
2016-11-22    26.308828
2016-11-23    26.174692
2016-11-24    26.240581
2016-11-25    26.306471
2016-11-26    26.289216
2016-11-27    26.271960
2016-11-28    26.254705
2016-11-29    26.228815
2016-11-30    26.007618
2016-12-01    25.765236
2016-12-02    25.861721
2016-12-03    25.799751
2016-12-04    25.737782
2016-12-05    25.675812
2016-12-06    25.873482
2016-12-07    26.127634
2016-12-08    26.384132
2016-12-09    26.814764
Name: Close, dtype: float64

【讨论】：

我相信它做到了 :) 感谢您的支持。