Possible deprecated use of .year in "aggregations" about polars HOT 3 CLOSED

No, the syntax works fine. I agree that the imports are a little ambiguous. The way the docs are set up is that the code snippets are actually .py files that are run, and the output is displayed in the docs.

When I said it needs updating, it's mainly because 1) the import isn't identified anywhere in the docs, which can be a source of confusion, and 2) the age calculation is incorrect. I don't know making it correct is worth it though, we can see what others would say.

from polars.

Yikes, yeah that needs updating, and the formula will not correctly compute the age. Quick help though: the date is a datetime.date object, which does have the year property:

from datetime import date  # see here
import polars as pl

def compute_age():
    return date(2021, 1, 1).year - pl.col("birthdate").dt.year()

df = pl.DataFrame({
    "birthdate": [date(1950, 1, 1)],
})

df.select(compute_age())

shape: (1, 1)
┌─────────┐
│ literal │
│ ---     │
│ i32     │
╞═════════╡
│ 71      │
└─────────┘

from polars.

Ah, thanks! That makes way more sense. Is the need for an update because of the overall syntax?

Also, I'm revisiting reading through docs after the holidays and just had this doc open so jogging my memory here. I see date and datetime are imported from datetime in the Column Selections subchapter/section. Would it make sense to add an explicit import call to datetime on this page as well?

from polars.