Repository navigation
Give Cp a noise variance without a nugget, seed from the scored parti… - #10
Merged
Merged
Conversation
…tion, keep clipped cells' area resolution_profile(): when the variogram gives no nugget (none fitted, a fit that did not converge, a range below the shortest lag, or `sac` a range alone) Cp was NA at every level, and anything reading Cp by default fell through to another criterion. Cp now takes Mallows' own noise variance, the residual mean square of the finest level whose cells hold two scored rows each on average, with a warning; attr(, "cp_noise") records which variance was used and print() shows it. select_resolution() returns $seeds, the centres of the partition the profile scored at the chosen level (every level's are kept in attr(, "centres")). get_voronoi_seeds(method = "kmeans") and voronoi_seeds_kmeans() return those centres when given the selection or the profile, so the Voronoi cells are the cells the criteria judged; the fresh k-means they ran put 5-16% of 300 points, and 26-29% of 2000, in a different cell. A count passed as a number still runs a fresh k-means. The Voronoi, grid and Delaunay builders dropped a clipped cell that st_intersection() returned as a GEOMETRYCOLLECTION (its area plus a line where it touches the boundary from outside), leaving its points with no cell: ten of thirty on an L-shaped boundary. Each collection is now reduced to its polygonal part. make_folds(nndm)'s min_train warning states the ECDF excess that fires it and gives the medians only as context, since they can differ the other way. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01AwT2vKb3g4LrJDB1CTbviN
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
…tion, keep clipped cells' area
resolution_profile(): when the variogram gives no nugget (none fitted, a fit that did not converge, a range below the shortest lag, or
saca range alone) Cp was NA at every level, and anything reading Cp by default fell through to another criterion. Cp now takes Mallows' own noise variance, the residual mean square of the finest level whose cells hold two scored rows each on average, with a warning; attr(, "cp_noise") records which variance was used and print() shows it.select_resolution() returns $seeds, the centres of the partition the profile scored at the chosen level (every level's are kept in attr(, "centres")). get_voronoi_seeds(method = "kmeans") and voronoi_seeds_kmeans() return those centres when given the selection or the profile, so the Voronoi cells are the cells the criteria judged; the fresh k-means they ran put 5-16% of 300 points, and 26-29% of 2000, in a different cell. A count passed as a number still runs a fresh k-means.
The Voronoi, grid and Delaunay builders dropped a clipped cell that st_intersection() returned as a GEOMETRYCOLLECTION (its area plus a line where it touches the boundary from outside), leaving its points with no cell: ten of thirty on an L-shaped boundary. Each collection is now reduced to its polygonal part.
make_folds(nndm)'s min_train warning states the ECDF excess that fires it and gives the medians only as context, since they can differ the other way.
Claude-Session: https://claude.ai/code/session_01AwT2vKb3g4LrJDB1CTbviN