New tool for visualization of time series and anomalies in streaming data

Abstract

A new sofware tool for simultaneous visualization of multiple time dependent signals, featuring a novel and highly useful combination of capabilities, and published as an open source solution is presented in this paper. The tool is designed to meet the needs of its users who expect lightweight, interactive & intuitive use and ease of deployment in current setups, including live monitoring systems with anomaly detection, highlighting and streaming data processing abilities. The functionality and motivation for our system is derived from various signal analysis applications, our research activities related to design and evaluation of neural network models, and from systems for continuous monitoring and anomaly detection (e.g. in IT or medical domains), which is demonstrated on simple use case examples.