https://github.com/karashiiro/lodestone-id-time

Data scraper, formula and reference implementation for the estimated creation time of a FFXIV character given its Lodestone ID.
https://github.com/karashiiro/lodestone-id-time

data ffxiv ffxiv-character lodestone

Last synced: 2 months ago
JSON representation

Data scraper, formula and reference implementation for the estimated creation time of a FFXIV character given its Lodestone ID.

• Host: GitHub
• URL: https://github.com/karashiiro/lodestone-id-time
• Owner: karashiiro
• License: mit
• Created: 2021-03-25T17:54:51.000Z (about 4 years ago)
• Default Branch: main
• Last Pushed: 2021-04-02T01:48:09.000Z (about 4 years ago)
• Last Synced: 2024-04-21T10:24:58.482Z (about 1 year ago)
• Topics: data, ffxiv, ffxiv-character, lodestone
• Language: Go
• Homepage: https://karashiiro.github.io/lodestone-when
• Size: 89.8 KB
• Stars: 2
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# lodestone-id-time

Data scraper, formula and reference implementation for the estimated creation time of a FFXIV character given its Lodestone ID. This formula can be used to get the creation times for any character, including ones with private achievement data. The creation time is estimated based on the first achievement time of a sample of characters. A web interface is also available [here](https://karashiiro.github.io/lodestone-when/).

The formula is subject to change in the future, as more data is collected.

## Reference implementation

A reference implementation is provided in [reference_impl.py](reference_impl.py).

The numbers used are generated from [estimate.py](estimate.py).

## Formula

![Formula for Excel timestamp calculation of character creation time from a Lodestone ID](formula.png)

## Graph

![Graph showing relationship between Lodestone ID and character creation time](graph.png)

Mean-squared error (Excel timestamp): 7422.162747

## Notes

* There's a marked slowdown around the beginning of 2020. If this were redone with

  machine learning, it might be smart to include a feature including lockdown data

  or some proxy for it.

* The 2020-Now part of the formula might be better approximated with another

  sinusoidal term, but it's unclear yet if that's the case.

* At the time this graph was constructed, there was a bug in my thread work distribution,

  which resulted in the gaps in the data. It has since been fixed.