ENH: add ignore_index to nlargest and nsmallest method

[erfannariman]

Describe the solution you'd like

Just like other methods where we select a subset or reorder, we have a ignore_index argument, to reset the index.

In [1]: import pandas as pd

In [2]: df = pd.DataFrame({"A": range(10), "B": range(10, 20)})

In [3]: df.nlargest(n=2, columns="A")
Out[3]: 
   A   B
9  9  19
8  8  18

In [4]: df.nlargest(n=2, columns="A", ignore_index=True)
Out[3]: 
   A   B
0  9  19
1  8  18

[alimcmaster1]

take

Thanks for the issue.

Lets get a +1 from a core developer before making an API changes. Saves you work if its not something that will be merged.

Not sure it makes sense to add this to just nlargest and nsmallest seems a bit selective.

[erfannariman]

Not sure it makes sense to add this to just nlargest and nsmallest seems a bit selective.

The reason I opened this issue is that we have ignore_index in drop_duplicates, sort_values, explode, sample and probably others as well.

[mroeschke]

My 2 cents; I'd personally prefer ignore_index to be removed in methods since it's equivalent to chaining reset_index(drop=True) correct?

[erfannariman]

My 2 cents; I'd personally prefer ignore_index to be removed in methods since it's equivalent to chaining reset_index(drop=True) correct?

I am not sure if that is consistent with the API design, since then you could also argue about as_index argument in groupby since you can achieve the same by chaining reset_index.

[mroeschke]

I am not sure if that is consistent with the API design, since then you could also argue about as_index argument in groupby since you can achieve the same by chaining reset_index.

xref #35860

[erfannariman]

I see, will close this for now, since there are discussions about deprecation and it looks it leans more in favour of actually doing so. So adding ignore_index does not make sense in this case.