Add tolerance for best_purity checks #119
Merged
Conversation
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
Member
|
Wow, good catch!! |
JackDunnNZ
added a commit
to JackDunnNZ/DecisionTree.jl
that referenced
this pull request
Jun 8, 2020
…` rather than an absolute tolerance
bensadeghi
added a commit
that referenced
this pull request
Jun 10, 2020
Fix #119 to work regardless of data scaling
giopaglia
pushed a commit
to giopaglia/ModalDecisionTrees.jl
that referenced
this pull request
Nov 27, 2022
Add tolerance for best_purity checks
giopaglia
pushed a commit
to giopaglia/ModalDecisionTrees.jl
that referenced
this pull request
Nov 27, 2022
…` rather than an absolute tolerance
giopaglia
pushed a commit
to giopaglia/ModalDecisionTrees.jl
that referenced
this pull request
Nov 27, 2022
Fix JuliaAI#119 to work regardless of data scaling
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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Outdated suggestions cannot be applied.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.
I found that I was getting different results for the same problem with the same RNG across different Julia versions (even after accounting for the RNG changes in Julia 1.5).
The differences seem to show up when there are multiple features with the same best split position, but because of differences between the Julia versions, the actual purity values are slightly different in the last digits, e.g.
9.893954468378139vs9.893954468378137, and this causes different features to be chosen on each version.This change simply adds a small tolerance to the purity check, which ensures that changes in the last digits won't affect which feature is selected as best.