A probability theoretic approach to drifting data in continuous time domains

TL;DR

This paper introduces a probability theoretical framework for formalizing and detecting drift in continuous time data, unifying existing notions and enabling new methods for drift detection and data decomposition.

Contribution

It provides a formal, stochastic dependency-based definition of drift in continuous time and develops a novel, efficient drift detection technique based on this framework.

Findings

New formalization of drift as stochastic dependency between data and time

Development of an efficient drift detection method

Method for decomposing data into drifting and non-drifting components

Abstract

The notion of drift refers to the phenomenon that the distribution, which is underlying the observed data, changes over time. Albeit many attempts were made to deal with drift, formal notions of drift are application-dependent and formulated in various degrees of abstraction and mathematical coherence. In this contribution, we provide a probability theoretical framework, that allows a formalization of drift in continuous time, which subsumes popular notions of drift. In particular, it sheds some light on common practice such as change-point detection or machine learning methodologies in the presence of drift. It gives rise to a new characterization of drift in terms of stochastic dependency between data and time. This particularly intuitive formalization enables us to design a new, efficient drift detection method. Further, it induces a technology, to decompose observed data into a drifting and a non-drifting part.