Benchmarking Change Detector Algorithms from Different Concept Drift Perspectives

The stream mining paradigm has become increasingly popular due to the vast number of algorithms and methodologies it provides address current challenges Internet Things (IoT) modern machine learning systems. Change detection algorithms, which focus on identifying drifts in data distribution during operation a solution, are crucial aspect this paradigm. However, selecting best change method for different types concept drift can be challenging. This work aimed provide benchmark four (EDDM, DDM, HDDMW, HDDMA) abrupt, gradual, incremental types. To shed light capacity possible trade-offs involved algorithm, we compare their capability, time, delay. experiments were carried out using synthetic datasets, where various attributes, such as size, amount drifts, duration controlled manipulated our generator stream. Our results show that HDDMW trade-off among all performance indicators, demonstrating superior consistency detecting abrupt but suboptimal time consumption limited ability detect drifts. outperforms other delay both gradual with an efficient performance.