Abstract:To design a successful Multiplayer Online Battle Arena (MOBA) game, the ratio of snowballing and comeback occurrences to all matches played must be maintained at a certain level to ensure its fairness and engagement. Although it is easy to identify these two types of occurrences, game developers often find it difficult to determine their causes and triggers with so many game design choices and game parameters involved. In addition, the huge amounts of MOBA game data are often heterogeneous, multi-dimensional and highly dynamic in terms of space and time, which poses special challenges for analysts. In this paper, we present a visual analytics system to help game designers find key events and game parameters resulting in snowballing or comeback occurrences in MOBA game data. We follow a user-centered design process developing the system with game analysts and testing with real data of a trial version MOBA game from NetEase Inc. We apply novel visualization techniques in conjunction with well-established ones to depict the evolution of players' positions, status and the occurrences of events. Our system can reveal players' strategies and performance throughout a single match and suggest patterns, e.g., specific player' actions and game events, that have led to the final occurrences. We further demonstrate a workflow of leveraging human analyzed patterns to improve the scalability and generality of match data analysis. Finally, we validate the usability of our system by proving the identified patterns are representative in snowballing or comeback matches in a one-month-long MOBA tournament dataset.

A Visual Analytics Approach for Understanding Reasons Behind Snowballing and Comeback in MOBA Games