Examples of Applying a Function in Python that takes in multiple inputs and using it to create a new column within an Ibis Dataframe?

[YCat33]

I currently have a calculate_date_diff() function that takes in two dates and then calculates the difference. I am using the mutate functionality to apply this function but the output seems to be the same regardless of the input:

date_calc = diff_types.mutate(string_date_diff=calculate_date_diff(diff_types.registration_date, diff_types.cc_open_date))

Would anyone be able to provide an example or guidance?

[gforsyth]

Hey @YCat33 -- can you post the contents of your diffing function? This should be possible as you've written it if the function is using the ibis API to operate on the two columns.

[gforsyth]

Hmm, DateDiff should be available for Trino for Ibis versions >= 7.1.0 -- which version of Ibis do you have installed?

[YCat33]

Do you have an example of how DateDiff would work? I am running 7.2.0

[gforsyth]

It should work by just subtracting. That said, interval types are a pain and there might be some issues there. I would recommend you follow @cpcloud's advice and try using .delta instead.

[gforsyth]

This is using the Trino backend:

[ins] In [13]: t
Out[13]: 
┏━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ cc_open_date ┃ date_reg   ┃
┡━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ string       │ date       │
├──────────────┼────────────┤
│ 2012-09-12   │ 2013-09-13 │
│ 2003-08-31   │ 2005-12-31 │
└──────────────┴────────────┘

[ins] In [14]: t.mutate(date_diff=_.date_reg.cast("date").delta(_.cc_open_date.cast("date"), "day"))
Out[14]: 
┏━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ cc_open_date ┃ date_reg   ┃ date_diff ┃
┡━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ string       │ date       │ int64     │
├──────────────┼────────────┼───────────┤
│ 2012-09-12   │ 2013-09-13 │       366 │
│ 2003-08-31   │ 2005-12-31 │       853 │
└──────────────┴────────────┴───────────┘

[ins] In [15]: def calc_date_diff(left, right):
          ...:     return left.cast("date").delta(right.cast("date"), "day")

[ins] In [16]: t.mutate(date_diff=calc_date_diff(_.date_reg, _.cc_open_date))
Out[16]: 
┏━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ cc_open_date ┃ date_reg   ┃ date_diff ┃
┡━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ string       │ date       │ int64     │
├──────────────┼────────────┼───────────┤
│ 2012-09-12   │ 2013-09-13 │       366 │
│ 2003-08-31   │ 2005-12-31 │       853 │
└──────────────┴────────────┴───────────┘