Narwhals support for CLV aggregation

[williambdean]

Description

Still a work in progress.

The LazyFrame like libraries will require a provided observation_end_date. However, that can be found outside of the

Still building out the functionality for the:

• remove first observation - mean with nans might differ in the backends

Related Issue

• Closes #
• Related to #

Checklist

• Checked that the pre-commit linting/style checks pass. Feel free to comment pre-commit.ci autofix to auto-fix.
• Included tests that prove the fix is effective or that the new feature works
• Added necessary documentation (docstrings and/or example notebooks) using numpydoc format.
• If you are a pro: each commit corresponds to a relevant logical change

---

📚 Documentation preview 📚: https://pymc-marketing--1809.org.readthedocs.build/en/1809/

[codecov]

Codecov Report

Attention: Patch coverage is 26.31579% with 14 lines in your changes missing coverage. Please review.

Project coverage is 40.45%. Comparing base (ca0c420) to head (f4804c8).

Files with missing lines	Patch %	Lines
pymc_marketing/clv/utils.py	26.31%	14 Missing ⚠️

❗ There is a different number of reports uploaded between BASE (ca0c420) and HEAD (f4804c8). Click for more details.

HEAD has 9 uploads less than BASE

Flag	BASE (ca0c420)	HEAD (f4804c8)
	22	13

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #1809       +/-   ##
===========================================
- Coverage   92.28%   40.45%   -51.84%     
===========================================
  Files          62       62               
  Lines        7469     7487       +18     
===========================================
- Hits         6893     3029     -3864     
- Misses        576     4458     +3882     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[juanitorduz]

OMG! 100% yes! 🥳

[ColtAllen]

OMG! 100% yes! 🥳

Indeed, thanks for starting this!

Do you think the current pandas functions should still be retained for a time even after this is merged? Also, it seems like the PR description in the original message requires more details.

[williambdean]

Do you think the current pandas functions should still be retained for a time even after this is merged? Also, it seems like the PR description in the original message requires more details.

I was just doing some comparisons of the two at the moment. However, I think that the new one should just take it's place.

[williambdean]

Maybe @ColtAllen is interested in taking this over?

[FBruzzesi]

Hey @williambdean 👋🏼 I just find out this PR 🔥 left a couple of comments that might help😇

[FBruzzesi]

I am not sure if this would work/is supported, but you might try to do:

Suggested change

	if observation_period_end is None:
	observation_period_end = transactions[datetime_col].cast(nw.Datetime).max()
	if observation_period_end is None:
	observation_period_end = pl.col("max").max()

to get the global max datetime value.

This might also help to avoid this requirement:

The LazyFrame like libraries will require a provided observation_end_date. However, that can be found outside of the

[FBruzzesi]

This is very tempting, but consider creating a new column between operations - I would be afraid that for pandas the casting happens multiple times instead of once

[FBruzzesi]

For some time now, it should be possible to pass an expression so that you can avoid the renaming down in the pipeline, but it's definitely more of a personal preference 😇

Suggested change

	.group_by(customer_id_col)
	.group_by(nw.col(customer_id_col).alias("customer_id"))