v0.4.14

This release fixes an issue with the new Tweet functionality.

In short, the logic checking for None values in the market.sim_day iterator was written incorrectly, causing only Orders with corresponding Tweets to get through to the output. This release improves the logic so all Orders get through, and only a portion of orders will have corresponding Tweets.

I also experimented with using LLMs to generate tweets, and while the results were great. It's just not tenable for now in terms of cost and speed given the volume. Will continue noodling on this though.

v0.4.13

This release adds a new data type and generated file: Tweets! These are semi-randomly generated based on Order Items and a Customer's assigned fan_level they receive on instantiation randomly. These are designed to be useful for ML/AI tasks like sentiment analysis or other unstructured text tasks.

v0.4.12

This release removes the pandas dependency from the package properly (in setup.py and pyproject.toml) and updates the typer dependency to the new package structure which uses [all] option by default. It also adds explicit data type casting to all values in the to_dict method of all the entities to ensure that data is being written properly to CSV now that there's no longer pandas as an intermediary to coerce data types.

v0.4.11

This release fixes a bug where subtotal + tax_paid != order_total. It also updates all the dependencies (numpy, typer, and Faker) and removes the dependency on pandas, opting for Python's built-in csv library instead, which turned out to be a great deal faster.

v0.4.10

This release adds a new store in Los Angeles and a HealthNut Customer persona that primarily purchases beverages in the morning in the summer. It re-adds the products CSV output of the Inventory, and does some internal refactoring and testing.

v0.4.9

Stable release. Publishing workflow is now fully operational. Next couple releases will be focused on refactoring code for robustness and clarity and adding test coverage. Once we have a better foundation we'll add some new data types and look at adding some more interesting behavior curves.

v0.4.8

Test release updating the project's name in setup.py and pyproject.toml to 'jafgen' to align with PyPI https://pypi.org/project/jafgen/, which looks to be a requirement for the automated release publishing flow.

v0.4.7

Test release with the proper url configured in the publish action (https://pypi.org/project/jafgen/)

v0.4.6

Test release updating pyproject.toml and setup.py with description and README information required for the PyPI publishing action, which will hopefully work now.

v0.4.5

Test release with publish workflow now running on Linux, which is required for the PyPI publish action. I missed one job last time. This should hopefully get the publish flow working and we'll have real releases from here.