Fig. 6: Ablation studies and scalability benchmark.
From: Data splitting to avoid information leakage with DataSAIL

Quality of DataSAIL splits (a) and runtime (b) as a function of the numbers of clusters K. c Effect of acceptable error margins ϵ and δ on split quality. d Runtimes of DataSAIL and other tools as a function of dataset size. In the legend, DC abbreviates DeepChem.