iPDP: On Partial Dependence Plots in Dynamic Modeling Scenarios
This addresses the need for explainable AI in dynamic modeling scenarios, such as applications with concept drift, but it is incremental as it extends an existing method (PDP) to a new context.
The paper tackles the problem of explaining dynamic machine learning models that adapt over time, which existing post-hoc explanation techniques like partial dependence plots (PDP) do not handle, by proposing iPDP, a model-agnostic framework that extends PDP to extract time-dependent feature effects in non-stationary environments, showing it approximates a time-dependent PDP variant and demonstrating its efficacy through experiments on real-world and synthetic data.
Post-hoc explanation techniques such as the well-established partial dependence plot (PDP), which investigates feature dependencies, are used in explainable artificial intelligence (XAI) to understand black-box machine learning models. While many real-world applications require dynamic models that constantly adapt over time and react to changes in the underlying distribution, XAI, so far, has primarily considered static learning environments, where models are trained in a batch mode and remain unchanged. We thus propose a novel model-agnostic XAI framework called incremental PDP (iPDP) that extends on the PDP to extract time-dependent feature effects in non-stationary learning environments. We formally analyze iPDP and show that it approximates a time-dependent variant of the PDP that properly reacts to real and virtual concept drift. The time-sensitivity of iPDP is controlled by a single smoothing parameter, which directly corresponds to the variance and the approximation error of iPDP in a static learning environment. We illustrate the efficacy of iPDP by showcasing an example application for drift detection and conducting multiple experiments on real-world and synthetic data sets and streams.