fbruzzesi/iso-week-date

Toolkit to work with str representing ISO Week date format

👋 Hi there, I'm Francesco

🦸‍♂️ About me

I am a data scientist with 5+ years of experience, a mathematician by training, a passionate learner and an open source enthusiast.

I have in depth experience working on tabular data, time series, Bayesian statistic and mathematical optimization. I tend to prefer simple, scalable and understandable solutions as opposed to over-complex models when not necessary to bring value.

💼 Employer

• Currently, I am a data scientist at HelloFresh🍋 in Berlin office.
• Previously I worked a senior data scientist at Edison⚡ one of the leading Italian electric utility company. Here I was involved maily on mid and down stream projects on tabular data, time series, anomaly detection, and optimization.

💻 OSS and side projects

When I have some spare time on weekends I maintain few projects:

• scikit-lego: since Sept. 2023 I am a maintainer of the project. The goal of the package is to allow to joyfully build new building blocks that are scikit-learn compatible. scikit-lego has 30k+ downloads a month
• ISO Week Date: iso-week-date is a toolkit to work with strings representing ISO Week date in two formats: week YYYY-WNN and week date YYYY-WNN-D.
• Timebasedcv: a python library that provides a cross validation strategies based on actual datetime values instead of indexes, of course compatible with scikit-learn cross validation API.
• Compclasses: a python utility library to simplify composition (over inheritance).
• Deczoo: a zoo for python decorators, a collection of decorators used on a daily basis.
• Webapps (Also open source):
  • ATP Stats is a webapp providing tennis analytics and insights in a more colorful and intuitive manner with respect to the official ATP Tour.
  • Tabular Data Annotator a webapp to bulk label tabular data from a csv.

Both webapps are developed using python dash framework and deployed on heroku I need to migrate them to another hosting platform.

📫 How to reach me

The best way it to find me on linkedin.

🔧 Tech stack

  

                

             

iso-week-date's Issues

(iso) quarter property

As in tableau, (iso) quarter should be quite easy to implement

• Q1 is W01 - W13
• Q2 is W14 - W26
• Q3 is W27 - W39
• Q4 is W40 -W52/W53

Access iso week days

Add:

• days property
• nth method

IsoWeekDate

Class to work with ISO Week date format "YYYY-WNN-D"

Additional parsers-converters

List of additional from_* and to_* methods to implement:

• Ordinal
• Ordinal date
• year-week (-day) numbers tuple

dataframe module isoweekdate functionalities

Extend dataframe modules to include isoweekdate functionalities mirroring those for isoweek

Move GHA to uv

Check for W53

Write a method to validate if a given year has 52 or 53 weeks

`property` to `cached_property`

Move year and week from property to cached_property.
Is it worth it?

Make validation step optional

Validation step is not needed in a series of operations (e.g. __add__) and just slows down the instance creation.

Allow `__add__`, `__sub__` with iterables

Expected

IsoWeek('2023-W01') + (1, 2, 3) 
# (IsoWeek('2023-W02'), IsoWeek('2023-W03'), IsoWeek('2023-W04'))

Docstring `Usage` section

Add Usage section in methods docstrings

regex enhancement

Improve regex logic to avoid additional validation after matching

validate series functionalities

pandas and polars functions to validate series conform to YYYY-WNN format

Docs (continuous) update

List of features missing from the docs:

• from_values and to_values week classes methods
• datetime_to_isoweekdate and isoweekdate_to_datetime (dataframe utils)

Pydantic type

Implementing type compatible with pydantic for IsoWeek and IsoWeekDate classes.

pandas & polars `is_in` and `contains` predicates

Polars utils

Add polars utils mimicking pandas utils module

Mike for multi-version docs

Mike : Manage multiple versions of your MkDocs-powered documentation via Git

Add typos check in pre-commit and gha

Adopt Narwhals

Description

Migrating to narwhals would:

• Allow to maintain one frame codebase instead of 2
• Enlarge the set of supported dataframe backend
• Possibly test narwhals features and reason about narwhals#267

Polars and Pandas extensions

Good opportunity to use pandas and polars api extensions modules (extending pandas and extending polars)

Proper version checking

Description

Currently checking for extra libraries (i.e. pandas, polars, and pydantic) is only done as

try:
    import module_name
except ImportError:
    raise ...

however, the module could be available but the version not satisfy the min requirement if not installed with iso-week-date directly.

We need a better way of checking available version(s).

Entrypoint?

Consider adding cli entrypoint for

• range
• weeksout

Pydantic integration

Possibility to make it into a Pydantic compatible type (cfr. pydantic-extra-types implementation)