Using complex operations in query method of pandas-CodePudding

Let say I have below code,

import pandas as pd
dat = pd.DataFrame({'col1' : [1,2,3,4,5,6,7,8,9,10], 'col2' : ['A', 'X', 'D', 'Y', 'A', 'D', 'Y', 'X', 'D', 'A']})
dat1 = pd.DataFrame({'col1' : [1,2,3,4,5,6,7,8,9,10], 'col2' : ['A', 'X', 'D', 'Y', 'A', 'D', 'Y', 'X', 'D', 'A']})
number = dat['col1'].values[dat.shape[0] - 3]
dat1.query("col1 == @number")

now instead of using @number in the last line, I want pass entire calculation dat['col1'].values[dat.shape[0] - 3] in the query method.

Is there any way to achieve this?

CodePudding user response：

What exactly is your use case? query does not support such complex expression.

You can however use loc with a callable, which enables you to use a dynamic expression in a pipeline:

dat1.loc[lambda d: d['col1'].eq(d['col1'].values[d.shape[0] - 3])]

output:

   col1 col2
7     8    X