查询数据框但仅将过滤器应用于列值不是 NaN 的行

Question

我有一个数据框 df:

num1 | count | count_min | count_max
a    | 10    | 5         | 10
b    | 15    | 6         | 11
c    | 3     | NaN       | NaN

我想过滤掉不在 count_min 和 count_max 之间的每个计数。

但如果 count_min/count_max 为 NaN，则应保留该行。

最终结果应该是：

num1 | count | count_min | count_max
a    | 10    | 5         | 10
c    | 3     | NaN       | NaN

所以我需要在查询中使用类似 if/else 的东西来检查 count_min/count_max 是否为 NaN，然后再应用过滤器。

如何使用如下查询语法实现此目的：

df = df.query("count >= count_min and count <= count_max")

?

Answer 1

使用Series.between and Series.isna:

In [4487]: df = df[df['count'].between(df.count_min, df.count_max) | (df.count_max.isna() | df.count_min.isna())]

In [4487]: df
Out[4487]: 
  num1  count  count_min  count_max
0    a     10        5.0       10.0
2    c      3        NaN        NaN

Answer 2

对于这种情况，您可以使用 np.where() 并将其用作过滤器：

df[np.where((df['count'].between(df['count_min'].values,df['count_max'].values)) | (df['count_min'].isna()) | (df['count_max'].isna()),True,False)]

Answer 3

像这样...

df = df[(df['count_min'].isna()) | (df['count_min'].isna()) | ((df['count'] >= df['count_min'] & (df['count'] <= df['count_max']))]

查询数据框但仅将过滤器应用于列值不是 NaN 的行

Query dataframe but apply filter only to rows where column value is not NaN

python

nan

filter

pandas