[英]Python2.7: FIlter out group from dataframe based on condition in groupby
I have a dataframe and I would like to filter the dataframe further to only include a group whose rows do not have a certain value in a column我有一个数据框,我想进一步过滤该数据框以仅包含其行在列中没有特定值的组
For eg, in the dataframe, since hamilton has an overtake in lap3 of his stint 1, I want to remove ALL of hamilton's stint 1 laptime records from the dataframe below.例如,在数据框中,由于汉密尔顿在他的第 1 阶段的第 3 圈超车,我想从下面的数据框中删除汉密尔顿的所有第 1 圈记录。
I thought of doing a groupby and then a get group,iterate through each row in the group, detect non-null value in the "clear lap?"我想到做一个groupby然后一个get group,遍历组中的每一行,在“clear lap”中检测非空值? column, and label "yes" in a new column for all rows in the groupby, then filter out the group.列,并在新列中为 groupby 中的所有行标记“是”,然后过滤掉该组。
Is there a faster way of subsetting the dataframe?有没有更快的方法来设置数据帧的子集?
Dataframe:数据框:
name driverRef stint tyre lap pos clear lap?
0 Australian Grand Prix vettel 1.0 Super soft 2 1 NaN
1 Australian Grand Prix vettel 1.0 Super soft 3 1 NaN
2 Australian Grand Prix vettel 1.0 Super soft 4 1 NaN
3 Australian Grand Prix ham 1.0 Super soft 2 3 NaN
4 Australian Grand Prix ham 1.0 Super soft 3 2 overtook
5 Australian Grand Prix ham 1.0 Super soft 4 2 NaN
I believe you need get all groups by filtering and then filter again by isin
:我相信您需要通过过滤获取所有组,然后通过isin
再次过滤:
Notice: Thank you, @Vivek Kalyanarangan for improvement by unique
.注意:谢谢@Vivek Kalyanarangan 通过unique
改进。
a = df.loc[df['clear lap?'].notnull(), 'driverRef'].unique()
print (a)
['ham']
df = df[~df['driverRef'].isin(a)]
print (df)
name driverRef stint tyre lap pos clear lap?
0 Australian Grand Prix vettel 1.0 Super soft 2 1 NaN
1 Australian Grand Prix vettel 1.0 Super soft 3 1 NaN
2 Australian Grand Prix vettel 1.0 Super soft 4 1 NaN
Another solution, slowier:另一个解决方案,速度较慢:
df = df[df['clear lap?'].isnull().groupby(df['driverRef']).transform('all')]
Or slowiest:或者最慢:
df = df.groupby('driverRef').filter(lambda x: x['clear lap?'].isnull().all())
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.