ENH: Use binary right shift as pipe operator

[JulianWgs]

Is your feature request related to a problem?

In R there is the pipe operator "%>%" which applies the function after the operator on the data before the operator. Pandas also follows this philosophy of method chaining. There is also a pipe function. Here is a good comparison of R vs pandas: https://stmorse.github.io/journal/tidyverse-style-pandas.html

Pandas is missing a stylish pipe operator.

Describe the solution you'd like

In [1]: import pandas as pd
   ...: import numpy as np
   ...:
   ...: def pipe(self, func):
   ...:     return func(self)
   ...:
   ...: pd.DataFrame.__rshift__ = pipe
   ...: pd.Series.__rshift__ = pipe

In [2]: df = pd.DataFrame({"A": [0, 1], "B": [2, 3]})
   ...: df
Out[2]:
   A  B
0  0  2
1  1  3

The old style. Execution from right to left or with pipe function.

In [3]: np.mean(np.sum(df))
Out[3]: 3.0

In [4]: df.pipe(np.sum).pipe(np.mean)
Out[4]: 3.0

With the pipe operator. Execution from left to right with pipe operator.

In [5]: df >> np.sum >> np.mean
Out[5]: 3.0

In [6]: df >> print
   A  B
0  0  2
1  1  3

I think the pipe operator makes it much more minimal.

API breaking implications

See above.

Describe alternatives you've considered

I picked the binary shift operator, because it is not currently used within pandas. I think it resembles a pipe the best.

Additional context

This does not work out of the box if the function takes additional argumements. One could use a lambda function, but that does not look clean. I think this is also the bigget downfall of this proposal.

[jreback]

This was already debated and was passed

-1 on this non python syntax.

I think there is a package that actually implements this on top of pandas if you want, but not for core pandas.

[JulianWgs]

Hey, thanks for the fast answer. Could you link that discussion? I would then gladly close the issue :)

Edit: I think I've found the package: https://github.com/dodger487/dplython

Edit2: Since there already is an implementation I will close this issue.

[jreback]

there are many just search for pipe
#27400 #26809