circa-clue 0.0.1

Causal Inference-based Root Cause Analysis

CIRCA

This project contains the code of baselines and simulation data generation for the KDD '22 paper, Causal Inference-Based Root Cause Analysis for Online Service Systems with Intervention Recognition. Experiment results can be found in figshare, where the code is corresponding to the commit 1522ddd7efd16db55e9f351fd70324501ce9134e.

Usage

This repository contains a Dockerfile to describe the necessary steps to setup the environment. To install this project as a package with pip, R package pcalg has to be installed manually.

Simulation Data Generation

python -m circa.experiment generate

Simulation Study

# Explore parameter combinations
python -m circa.experiment --max-workers 16 --model-params params-sim-tune.json tune
# Explore all the datasets with pre-defined parameters
python -m circa.experiment --model-params params-sim-run.json run
# Robustness evaluation
python -m circa.experiment robustness

Execute Rscript img/draw.sim.R to produce summaries under img/output.

• params-sim-run.json is created according to img/output/best-sim-tuning.tex
• To create parameter template, execute the following command

python -m circa.experiment params > default.json