In this paper, we present GameDepot, a visual analytics system designed to enable interactive analysis of performance testing logs for mobile games. Due to the emergence of computational-heavy mobile games, the performance testing of mobile games has become a crucial practice to strike a balance between user experience, battery life, and thermal management. However, policymakers in device manufacturers face challenges in understanding the complex performance testing logs generated for different game and device combinations. To address this issue, we conduct a design study with ten domain experts and develop GameDepot, a visual analytics system with six tightly coordinated views that provide an overview of thousands of testing sessions and facilitate the identification of abnormal events, such as excessive throttling. We implement a data modeling pipeline based on Long Short Term Memory (LSTM) to support advanced data manipulation operations that are needed in practice, such as measuring the similarity between devices and or predicting the performance of an unseen combination of games and devices. Our evaluation demonstrates that GameDepot not only supports the seven essential tasks we identified during the design study but also facilitates the identification of abnormal sessions.
CITATION STYLE
Jang, D., & Jo, J. (2023). GameDepot: A Visual Analytics System for Mobile Game Performance Testing. IEEE Access, 11, 83251–83263. https://doi.org/10.1109/ACCESS.2023.3301775
Mendeley helps you to discover research relevant for your work.