Robust PCA for Anomaly Detection and Data Imputation in Seasonal Time Series

TL;DR

This paper introduces a robust PCA framework tailored for anomaly detection and data imputation in seasonal time series, including an online algorithm for streaming data and empirical validation of its effectiveness.

Contribution

It presents a novel online RPCA algorithm for seasonal time series and demonstrates its advantages over existing methods through empirical comparisons.

Findings

Effective anomaly detection in seasonal data.

Successful data imputation in practical scenarios.

Online algorithm scales to large datasets.

Abstract

We propose a robust principal component analysis (RPCA) framework to recover low-rank and sparse matrices from temporal observations. We develop an online version of the batch temporal algorithm in order to process larger datasets or streaming data. We empirically compare the proposed approaches with different RPCA frameworks and show their effectiveness in practical situations.