sktime/sktime

[ENH] sliding window (z or quantile) score based anomaly detection

Open

#6,811 opened on Jul 22, 2024

View on GitHub
 (7 comments) (1 reaction) (0 assignees)Python (1,192 forks)batch import
enhancementgood first issuemodule:detection

Repository metrics

Stars
 (7,162 stars)
PR merge metrics
 (Avg merge 26d 10h) (86 merged PRs in 30d)

Description

Two algorithms that we could implement for starter in the sliding windo detection family:

  • sliding window with parameters related to window (like in pandas). Get quantile or z score based on last window, and threshold the score.
  • fit-update-predict on expanding or rolling window splitter with an arbitrary forecaster. Get forecasts, and threshold residuals. Or get proba forecasts, and compute quantile score based on those.

The first is a special case of the second, where we make a naive distributional forecast, but I think it's worth having the first separately implemented for two reasons:

  • it is a common subcase, and not obvious how to obtain it as a special case of the second
  • we explore implementation issues that arise from using sliding windows etc

Contributor guide

Research direction
Explore sktime's existing sliding window and forecasting infrastructure. Understand the base classes for detectors and forecasters. Implement naive distributional forecasting for the first algorithm and use rolling/expanding window splitters for the second.
Tech stack
python
Domain
backend
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.
3-5 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
Active
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
Clear
Prerequisites
PythonGit
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
55