Simple Root Cause Analysis by Separable Likelihoods

TL;DR

This paper introduces a Bayesian framework for simple and interpretable Root Cause Analysis (RCA) that leverages separable likelihoods, making it practical for industrial anomaly detection tasks.

Contribution

It proposes a novel RCA method based on separable likelihoods within the Bayesian regime, applicable to models like Multinomial and Naive Bayes, with validation on real-world data.

Findings

Framework is effective for web server error logs

Applicable to important base models like Multinomial and Naive Bayes

Provides a balance between accuracy and interpretability

Abstract

Root Cause Analysis for Anomalies is challenging because of the trade-off between the accuracy and its explanatory friendliness, required for industrial applications. In this paper we propose a framework for simple and friendly RCA within the Bayesian regime under certain restrictions (that Hessian at the mode is diagonal, here referred to as \emph{separability}) imposed on the predictive posterior. We show that this assumption is satisfied for important base models, including Multinomal, Dirichlet-Multinomial and Naive Bayes. To demonstrate the usefulness of the framework, we embed it into the Bayesian Net and validate on web server error logs (real world data set).