Hi! I work with ad-hoc datasets a lot and Chrome is my main lens into them. I decided to build an extension that shades the comparable columns of those tables the way a spreadsheet's conditional formatting would.
Quickly figured out that there's a problem with just applying "big numbers = green, small numbers = red" uniformly, though. On my fantasy football team page, "number of fumbles" should show green when a player has few / red when they have many.
When TypesafeAI's Jev decision model was released, it seemed like a really good fit for being able to solve this sort of problem. With some context (like "does `cost` here seem like a good thing or a bad thing, given the title of the table and adjacent columns), it is great (and quick!) at determining expected order.
Was already doing the work of parsing numbers from cells, so things like "$50-60k" becomes comparable and so do cells with adjacent text (like "32.4%
Xhigh effort").
It's all open source[0] and you can BYOK (for Jev or Clef) or use it for free (with some healthy limits)
Hi! I work with ad-hoc datasets a lot and Chrome is my main lens into them. I decided to build an extension that shades the comparable columns of those tables the way a spreadsheet's conditional formatting would.
Quickly figured out that there's a problem with just applying "big numbers = green, small numbers = red" uniformly, though. On my fantasy football team page, "number of fumbles" should show green when a player has few / red when they have many.
When TypesafeAI's Jev decision model was released, it seemed like a really good fit for being able to solve this sort of problem. With some context (like "does `cost` here seem like a good thing or a bad thing, given the title of the table and adjacent columns), it is great (and quick!) at determining expected order.
Was already doing the work of parsing numbers from cells, so things like "$50-60k" becomes comparable and so do cells with adjacent text (like "32.4% Xhigh effort").
It's all open source[0] and you can BYOK (for Jev or Clef) or use it for free (with some healthy limits)
I hope you like it!
[0]: https://github.com/philrenaud/practical-data