yzhao062/pyod

ECOD and COPOD `explain_outlier` on new data?

Open

#408 opened on May 17, 2022

View on GitHub
 (2 comments) (0 reactions) (0 assignees)Python (7,762 stars) (1,308 forks)batch import
enhancementhelp wanted

Description

Hi there,

Really like the python package, have been using it for a while. I was wondering if it was possible for explain_outlier method in ECOD and COPOD to be adapted as to accept new data not used during training?

Or at least to return the scores per feature given a event, as I don't need the graphic just to interpret why something was flagged as anomalous/has such a high anomaly score.

Cheers

Contributor guide

Tech stack
python
Domain
machine learningdata
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-2 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
clear
Prerequisites
Pythonpyod libraryanomaly detection basics
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
40
Research direction
Examine the current implementation of explain outlier in ECOD and COPOD (likely in pyod/models/ecod.py and pyod/models/copod.py) to understand how it computes per feature scores using training data. Then design modifications to accept new data at inference time, possibly by caching necessary statistics from training. Also consider separating the score computation from the plotting function to allow users to retrieve scores programmatically. Check existing documentation and ensure backward compatibility.