This dataset contains the results (in .csv format) of experiments performed using the code from the paper "Minimizing Overprocessing Waste via Predictive Activity Ordering" (https://doi.org/10.1007/978-3-319-39696-5_12). Each file corresponds to an execution of the entire training & evaluation pipeline, each time with different random subsets of the full dataset. They contain the number of knockout checks performed with 3 different approaches (variable rejection probabilities, constant rejection probabilities, and random ordering).