A better understanding of the system and the user’s analytic process give rise to opportunities to create adaptive systems. Such approaches are prominent in the existing literature and seek to improve the usability and performance of a visualization system, or the collaborative potential of the visual analytics tool. The body of prior work includes a wide variety of systems that recommend visualizations based on inferred tasks, provide guidance for a given interface, or prefetch data to improve system performance. For example, Gotz and Wen proposed behavior-driven visualization recommendation that infers a user’s task in real-time and suggests an alternative visualization that might support the task better. A similar approach was adopted by Mutlu et al. by adapting visualization recommendations to the users’ preferences. Fan et al. trained a convolutional neural network on interaction data to create a faster and more accurate scatter plot brushing tool. By analyzing real-time interactions, Battle et al. demonstrated that incorporating provenance data into the prefetching pipeline improved system latency by 430%. To explore event sequence predictions, Guo et al. preserve and aggregate records by their top prediction. In order to achieve a higher acceptance rate of the predictions, they showed multiple predictions and let the user choose.