In carrying out its banking supervision tasks as part of the Single Supervisory Mechanism (SSM), the European Central Bank (ECB) collects and disseminates data on significant and less significant institutions. To ensure harmonised supervisory reporting standards, the data are represented through the European Banking Authority's data point model, which defines all the relevant business concepts and the validation rules. For the purpose of data quality assurance and assessment, ECB experts may implement additional plausibility checks on the data. The ECB is constantly seeking ways to improve these plausibility checks in order to detect suspicious or erroneous values and to provide high-quality data for the SSM. In this paper we describe a data-driven approach, based on machine learning, for discovering new plausibility checks. Specifically, the approach makes use of large amounts of historical data to identify patterns in past observations. The patterns of interest correspond to latent and potentially non-linear relationships in the data, which serve as a basis for defining new checks. We show that this approach can be used to detect relevant patterns and that these patterns are suitable for discovering anomalies in the data. We also illustrate how such patterns are used by business experts to refine their data quality framework. We finally provide suggestions for potential further work that could be carried out to improve technical performance as well as prediction quality.