我有不规则的时间序列数据，应该是每秒钟50个样本，但是有些时候它有10个样本，有些时候是一秒100个样本。我想在同一秒内将10个样本插入到50个样本中，并在一秒内删除50个样本以上的任何样本
为简单起见，我将正常采样率描述为5
TIMESTAMP X Y Z
2017-10-31 11:11:22 -1.451385 -9.44754 -0.876785
2017-10-31 11:11:22 -1.435913 -9.628448 -0.914871
2017-10-31 11:11:22 -1.397827 -9.915283 -0.976761
2017-10-31 11:11:23 -1.234772 -10.162842 -0.808945
2017-10-31 11:11:23 -1.234772 -10.003357 -0.637558
2017-10-31 11:11:23 -1.316895 -9.918854 -0.290024
2017-10-31 11:11:23 -1.362122 -10.08786 0.070602
2017-10-31 11:11:23 -1.28595 -10.366364 0.276505
2017-10-31 11:11:24 -1.164551 -10.548462 0.472885
2017-10-31 11:11:24 -1.150269 -10.774597 0.82756
2017-10-31 11:11:24 -1.153839 -10.895996 1.163193
2017-10-31 11:11:24 -1.165741 -10.732941 1.304825
2017-10-31 11:11:24 -1.249054 -10.367554 1.07988
2017-10-31 11:11:24 -1.314514 -9.936707 0.935867
2017-10-31 11:11:24 -1.393066 -9.653442 0.768051
2017-10-31 11:11:25 -1.583496 -9.25235 0.69664
2017-10-31 11:11:25 -1.944122 -9.070251 0.344345
2017-10-31 11:11:25 -2.358307 -9.057159 -0.031754
2017-10-31 11:11:25 -2.477325 -9.035736 -0.39119
我希望它是
TIMESTAMP X Y Z
2017-10-31 11:11:22 -1.451385 -9.44754 -0.876785
2017-10-31 11:11:22 -1.435913 -9.628448 -0.914871
2017-10-31 11:11:22 -1.397827 -9.915283 -0.976761
2017-11-01 11:11:22 NaN NaN NaN
2017-11-02 11:11:22 NaN NaN NaN
2017-10-31 11:11:23 -1.234772 -10.162842 -0.808945
2017-10-31 11:11:23 -1.234772 -10.003357 -0.637558
2017-10-31 11:11:23 -1.316895 -9.918854 -0.290024
2017-10-31 11:11:23 -1.362122 -10.08786 0.070602
2017-10-31 11:11:23 -1.28595 -10.366364 0.276505
2017-10-31 11:11:24 -1.164551 -10.548462 0.472885
2017-10-31 11:11:24 -1.150269 -10.774597 0.82756
2017-10-31 11:11:24 -1.153839 -10.895996 1.163193
2017-10-31 11:11:24 -1.165741 -10.732941 1.304825
2017-10-31 11:11:24 -1.249054 -10.367554 1.07988
2017-10-31 11:11:25 -1.583496 -9.25235 0.69664
2017-10-31 11:11:25 -1.944122 -9.070251 0.344345
2017-10-31 11:11:25 -2.358307 -9.057159 -0.031754
2017-10-31 11:11:25 -2.477325 -9.035736 -0.39119
2017-11-01 11:11:25 NaN NaN NaN
df.resample('10ms').first().ffill()
它做了一些插值，但是没有提供我所需要的