Exponentially weighted moving average charts for detecting concept drift

Abstract

Classifying streaming data requires the development of methods which are computationally efficient and able to cope with changes in the underlying distribution of the stream, a phenomenon known in the literature as concept drift. We propose a new method for detecting concept drift which uses an Exponentially Weighted Moving Average (EWMA) chart to monitor the misclassification rate of an streaming classifier. Our approach is modular and can hence be run in parallel with any underlying classifier to provide an additional layer of concept drift detection. Moreover our method is computationally efficient with overhead O(1) and works in a fully online manner with no need to store data points in memory. Unlike many existing approaches to concept drift detection, our method allows the rate of false positive detections to be controlled and kept constant over time.

An Efficient Concept Drift Detection Method for Streaming Data under Limited Labeling

“…(1) DDM [11], EDDM [12], EWMA [13]: Drift detection methods are used along with a incrementally updatable classifier which require all data samples to be labeled. (2) uMD (using Margin Density) [7], CDBD (Confidence Distribution Batch Detection) [6]: Using SVM as a classifier, it performs drift detection on unlabeled data streams.…”

“…Early Drift Detection Method (EDDM) [12] detects changes by analyzing the distance between two errors. Exponentially Weighted Moving Average (EWMA) [13] method uses exponentially weighted moving average chart Copyright c 2017 The Institute of Electronics, Information and Communication Engineers to monitor the misclassification rate of the classifier.…”

Yarn-Dyed Fabric Defect Detection Based On Autocorrelation Function And GLCM

“…It is difficult to determine pattern period according to the original curves directly. In order to compute the period of the fabric, data smooth processing should be performed to get rid of the local peaks and valleys [12]. Locally Weighted Scatterplot Smoothing (LOWESS) [13] is used, and the window widths are 50 for both directions.…”

Pitfalls in Benchmarking Data Stream Classification and How to Avoid Them

“…Although there is much research in the data stream literature on detecting concept drift and adapting to it over time [10,17,21], most work on stream classification assumes that data is distributed not identically, but still independently. Except for our brief technical report [24], we are not aware of any work in data stream classification discussing what effects a temporal dependence can have on evaluation.…”

“…The Electricity dataset due to [15] is a popular benchmark for testing adaptive classifiers. It has been used in over 40 concept drift experiments 5 , for instance, [10,17,6,21]. The Electricity Dataset was collected from the Australian New South Wales Electricity Market.…”