Author(s): Karen Schulz; Andre Niemann; Thorsten Mietzel
Linked Author(s):
Keywords: Water resources management anomaly detection error detection unsupervised timeseries
Abstract: Data quality is fundamental to innovative, data-driven applications. Conventional data quality control methods often operate unsupervised and without the need for reference data to be considered reliable. Machine learning techniques have the potential to significantly improve data quality control in the water sector. A key challenge with current machine learning methods is their reliance on reference data to build trust, which can be difficult to obtain. In this work, we propose three complementary concepts to tackle this issue through model evaluation, each of which is applied to a real-world example dataset derived from low-cost water level sensors. Nevertheless, depending on the application scenario, the question remains as to whether these concepts are sufficient for deploying models without having reference sensors.
Year: 2025