Skip to content

Latest commit

 

History

History
41 lines (22 loc) · 1.47 KB

CHANGELOG.md

File metadata and controls

41 lines (22 loc) · 1.47 KB
Raw

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog

[Unreleased]

[0.8.5] - 2020-07-24

Changed

  • PointDensity.cdf() now does not interpolate and instead returns the cdf to the nearest grid point.
  • The default grid density for PoinDensity distributions is now 400 (from 200)

[0.8.4] - 2020-07-15

Changed

  • All arguments passed to Scales are now floats and all Scale fields are assumed to be floats

[0.8.3] - 2020-07-12

Added

  • PointDensity() distribution. We now primarily operate on point densities located in the center of what used to be the bins in our histogram.
  • "denorm_xs_only" option on PointDensity egress methods. This returns normalized probability densities but on a denormalized x-axis, as Metaculus displays.
  • support for LogScale and Metaculus/Elicit LogScale questions

Changed

  • We operate on 200 point densities at the center of bins evenly spaced from 0 to 1 on a normalized scale. If we are passed in anything besides this in from_pairs we interpolate to get 200 points placed in this manner.
  • We always operate on normalized points and normalized densities internally, using the normed_xs and normed_densities fields respectively. We denormalize the density before returning in PDF.
  • PointDensity.cdf(x) returns the cdf up to x (not the bin before x, as previously)

Removed

  • the Histrogram() distribution