automl / automl/ParameterImportance

An error found in influence-models

Open
#83 3 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
78
Forks
19
PR merge metrics
No merged PRs in 30d

Description

In influence_models.py line 115, the code is:
`tmp_used = used[:i] + used[i:]`
This line doesn't remove the ith element in used, maybe what you really want is sth like:
`tmp_used = used[:i] + used[i+1:]`
and also need to check whether i+1 is out of range.
For now, because 'tmp_used' is the same as 'used', backward pass actually did nothing.

Contributor guide

No contributing guide indexed for this repository

Research direction

Start at influence_models.py line 115 and inspect the backward pass around the tmp_used assignment. Verify how the selected index is excluded and how the final index is handled. Done means the backward pass no longer operates on an unchanged used list and remains safe at the list boundary.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
2/5
Estimated time
1-3 hours
Activity status
Stale
Clarity
Clearly specified
Newbie friendliness
52/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.