Inquiry About Inclusion of Comprehensive RCA Methods in Our RCAEval Project

[beifangqiuer]

Dear Phamquiluan
I am reaching out to inquire whether our project currently encompasses all the Root Cause Analysis (RCA) methods that are prevalent in both academic and industrial settings. Specifically, I am also interested in the following methods:

•   MicroGBPM
  

•   CORAL
  

•   REASON
  

•   Microscope
  

•   TON18, MonitorRank
  

Furthermore, I would like to know if there are any plans to update our project to include additional RCA methods. Incorporating a broader range of methodologies could significantly enhance our project's robustness and applicability.

Thank you for considering this request. I appreciate your guidance and look forward to your insights.

Warm regards,

[phamquiluan]

Thanks @beifangqiuer for the question.

whether our project currently encompasses all the Root Cause Analysis (RCA) methods that are
prevalent in both academic and industrial settings

We have many baselines that are published at top academic venues, e.g., RUN, CausalRCA, CIRCA, RCD, MicroCause, EasyRCA, MSCRED, BARO, 𝜖-Diagnosis, TraceRCA, MicroRank.

For industrial baselines, (to the best of our knowledge) PyRCA library from Saleforces offers 𝜖-Diagnosis, RCD, and HT (i.e., CIRCA), which is limited compared to our library.

I would like to know if there are any plans to update our project to include additional RCA methods.

I personally would love to integrate these methods into RCAEval and will be ready to review any Pull Requests from contributors. For my current research plan, I'm focusing more on the development of new RCA methods with more capacibility to diagnose more complex faults (e.g., code-level faults).

[beifangqiuer]

Thank you for your response. Let's work together to make progress in the field of causal inference!

[beifangqiuer]

I am writing to seek clarification regarding the compatibility of certain algorithms with Python 3.10. In the code snippet I am working with, it appears that the algorithms ht, rcd, and mmrcd are not imported when the environment is detected as Python 3.10:

if is_py310():
from RCAEval.e2e import (
baro,
causalrca,
circa,
cloudranger,
cmlp_pagerank,
dummy,
e_diagnosis,
easyrca,
fci_pagerank,
fci_randomwalk,
ges_pagerank,
granger_pagerank,
granger_randomwalk,
lingam_pagerank,
lingam_randomwalk,
micro_diag,
microcause,
microrank,
mscred,
nsigma,
ntlr_pagerank,
ntlr_randomwalk,
pc_pagerank,
pc_randomwalk,
run,
tracerca,
)
else:
from RCAEval.e2e import dummy, e_diagnosis, ht, rcd, mmrcd
Could you please confirm whether this implies that the ht, rcd, and mmrcd algorithms cannot be executed in a Python 3.10 environment? If so, are there any recommended alternatives or plans to support these algorithms in Python 3.10?

Thank you very much for your assistance.

Best regards,

[phamquiluan]

Hi @beifangqiuer, the implementation of rcd, mmrcd is tied to python 3.8 since they modify many dependencies. https://github.com/azamikram/rcd.

The ht method is from PyRCA package (salesforce/PyRCA#61). They depend on a very large java deps. So, I don't want to include them in the default mode.

[phamquiluan]

Please refer to https://github.com/phamquiluan/RCAEval/blob/main/docs/SETUP.md to see how we could use RCD, and HT method @beifangqiuer

[beifangqiuer]

Understood, thank you!

[beifangqiuer]

Dear Phamquiluan

In the field of root cause analysis, it should also include association rule mining, such as the Apriori algorithm, and heuristic search, such as the RiskLoc algorithm. Could you please update the project to make it more comprehensive? Thank you sincerely.

Best regards,

[phamquiluan]

Hi @beifangqiuer, could you please send me the paper and/or the source code of these algorithms. I'm happy to have a look and see if I can integrate these methods into our framework.