Project - Process Predictions with Process Knowledge: Considering Data Quality Issues
There has been a growing interest in using contemporary deep learning algorithms by applying sophisticated architectures to predict business process behaviour, such as the next event of a running case. However, existing methods often overlook the underlying characteristics of the process that result in data complexities. For example, data collected from different systems could result in data leakage, class imbalance or class overlap. This work aims to motivate the need to identify the data characteristics using the knowledge of the process. First, we will measure the performance of existing state of the art models and highlight the prediction performance by considering the quality of data. Finally, we will propose an approach that identifies and addresses complexities in the input process data through this work.
In progress