从 Pandas 系列中删除零行答案

【问题标题】：Remove rows of zeros from a Pandas series从 Pandas 系列中删除零行
【发布时间】：2015-11-11 01:46:23
【问题描述】：

我有一个编号Pandas 系列，其中包含按日期索引的 601 行，如下所示。这些值在某一点之前为零，之后所有值都非零。这一点因每个系列而异，但我想要一种方法来删除值为零的所有行，同时保持日期索引的完整性。

Name: users, dtype: float64 dates
2015-08-17 14:29:59-04:00    18
2015-08-16 14:29:59-04:00     3
2015-08-15 14:29:59-04:00    11
2015-08-14 14:29:59-04:00    12
2015-08-13 14:29:59-04:00     8
2015-08-12 14:29:59-04:00    10
2015-08-11 14:29:59-04:00     6
2015-08-10 14:29:59-04:00     6
2015-08-09 14:29:59-04:00     7
2015-08-08 14:29:59-04:00     7
2015-08-07 14:29:59-04:00    13
2015-08-06 14:29:59-04:00    16
2015-08-05 14:29:59-04:00    12
2015-08-04 14:29:59-04:00    14
2015-08-03 14:29:59-04:00     5
2015-08-02 14:29:59-04:00     5
2015-08-01 14:29:59-04:00     8
2015-07-31 14:29:59-04:00     6
2015-07-30 14:29:59-04:00     7
2015-07-29 14:29:59-04:00     9
2015-07-28 14:29:59-04:00     7
2015-07-27 14:29:59-04:00     5
2015-07-26 14:29:59-04:00     4
2015-07-25 14:29:59-04:00     8
2015-07-24 14:29:59-04:00     8
2015-07-23 14:29:59-04:00     8
2015-07-22 14:29:59-04:00     9
2015-07-21 14:29:59-04:00     5
2015-07-20 14:29:59-04:00     7
2015-07-19 14:29:59-04:00     6
                             ..
2014-01-23 13:29:59-05:00     0
2014-01-22 13:29:59-05:00     0
2014-01-21 13:29:59-05:00     0
2014-01-20 13:29:59-05:00     0
2014-01-19 13:29:59-05:00     0
2014-01-18 13:29:59-05:00     0
2014-01-17 13:29:59-05:00     0
2014-01-16 13:29:59-05:00     0
2014-01-15 13:29:59-05:00     0
2014-01-14 13:29:59-05:00     0
2014-01-13 13:29:59-05:00     0
2014-01-12 13:29:59-05:00     0
2014-01-11 13:29:59-05:00     0
2014-01-10 13:29:59-05:00     0
2014-01-09 13:29:59-05:00     0
2014-01-08 13:29:59-05:00     0
2014-01-07 13:29:59-05:00     0
2014-01-06 13:29:59-05:00     0
2014-01-05 13:29:59-05:00     0
2014-01-04 13:29:59-05:00     0
2014-01-03 13:29:59-05:00     0
2014-01-02 13:29:59-05:00     0
2014-01-01 13:29:59-05:00     0
2013-12-31 13:29:59-05:00     0
2013-12-30 13:29:59-05:00     0
2013-12-29 13:29:59-05:00     0
2013-12-28 13:29:59-05:00     0
2013-12-27 13:29:59-05:00     0
2013-12-26 13:29:59-05:00     0
2013-12-25 13:29:59-05:00     0

【问题讨论】：

标签： python pandas series

【解决方案1】：

只需过滤掉它们：

users[users!=0]

这也将保留您的索引

或者

users[users > 0]

如果你追求的是正值：

In [38]:
s[s>0]

Out[38]:
2015-08-17 18:29:59    18
2015-08-16 18:29:59     3
2015-08-15 18:29:59    11
2015-08-14 18:29:59    12
2015-08-13 18:29:59     8
2015-08-12 18:29:59    10
2015-08-11 18:29:59     6
2015-08-10 18:29:59     6
2015-08-09 18:29:59     7
2015-08-08 18:29:59     7
2015-08-07 18:29:59    13
2015-08-06 18:29:59    16
2015-08-05 18:29:59    12
2015-08-04 18:29:59    14
2015-08-03 18:29:59     5
2015-08-02 18:29:59     5
2015-08-01 18:29:59     8
2015-07-31 18:29:59     6
2015-07-30 18:29:59     7
2015-07-29 18:29:59     9
2015-07-28 18:29:59     7
2015-07-27 18:29:59     5
2015-07-26 18:29:59     4
2015-07-25 18:29:59     8
2015-07-24 18:29:59     8
2015-07-23 18:29:59     8
2015-07-22 18:29:59     9
2015-07-21 18:29:59     5
2015-07-20 18:29:59     7
2015-07-19 18:29:59     6
Name: 1, dtype: int64

【讨论】：

是否有一种流畅的方式来做到这一点（方法链，例如作为表达式末尾的过滤器）？

【解决方案2】：

如果ds 是你DataSeries：ds!=0 将返回一个布尔向量，该向量的值不为零。

ds[ds!=0] 是行，保留索引

请注意，缺失值 (NaN) 不会被过滤。

要过滤两者，请使用：ds[(ds!=0)&(pd.isnull(ds))]

【讨论】：

ds[(ds!=0)&(pd.isnull(ds))] 没有帮助删除 value = NaN 的行。