Concept Drift Monitoring and Diagnostics of Supervised Learning Models via Score Vectors

Abstract

Supervised learning models are one of the most fundamental classes of models. Viewing supervised learning from a probabilistic perspective, the set of training data to which the model is fitted is usually assumed to follow a stationary distribution. However, this stationarity assumption is often violated in a phenomenon called concept drift, which refers to changes over time in the predictive relationship between covariates X and a response variable Y and can render trained models suboptimal or obsolete. We develop a comprehensive and computationally efficient framework for detecting, monitoring, and diagnosing concept drift. Specifically, we monitor the Fisher score vector, defined as the gradient of the log-likelihood for the fitted model, using a form of multivariate exponentially weighted moving average, which monitors for general changes in the mean of a random vector. In spite of the substantial performance advantages that we demonstrate over popular errorbased methods, a score-based approach has not been previously considered for concept drift monitoring. Advantages of the proposed score-based framework include applicability to any parametric model, more powerful detection of changes as shown in theory and experiments, and inherent diagnostic capabilities for helping to identify the nature of the changes.

Concept Drift and Covariate Shift Detection Ensemble with Lagged Labels

“…It is worth pointing out that [18,19] also transformed the gradient by the inverse Fisher information matrix in a manner that is quite similar, but not identical, to our method. Namely, [18,19] monitored the components of the 'decoupled' score vector (i.e. the gradient transformed by the inverse Fisher information matrix).…”

“…Gradient changes: Proposed in [18,19], the changes of gradients can also be used to tackle the drift detection task, i.e. the larger the gradient changes, the more likely the data has changed.…”