Understanding Human Skill Acquisition through Statistical Modelling of Big Data in eSports

专业知识和技能在社会中发挥着越来越重要的作用，因为经济日益以知识为导向，涉及到快速的技术变革和越来越长的工作寿命。游戏玩家在几个小时、几天甚至几年的练习中发展出深厚的技能，产生了关于游戏和技能发展的海量数据，可以轻松而不引人注目地记录下来。其结果是一个无与伦比的机会来调查游戏的哪些方面决定了专家技能，以及它是如何获得的。拟议的博士学位的主题是测试技能获得的理论，并通过对电子竞技游戏中自然出现的数据的统计询问来表征游戏中的专家技能。这项研究将在约克大学的安德斯·德拉钦教授和谢菲尔德大学的汤姆·斯塔福德博士的共同监督下进行，主题是利用电子竞技数据理解人类心理。拟议的研究与电子竞技行业的现有链接(即DC实验室)、现有的专业知识(例如Alex Wade教授)以及IGGI通过分析主要在线游戏(例如英雄联盟、DOTA 2、命运)的玩家数据对表现和技能获取进行的持续分析非常吻合。技能获取的认知科学具有广泛的适用性，对电子竞技行业具有实际意义。开发可预测高水平表现的技能衡量标准，可以让专业团队更有效地识别人才，并允许游戏开发人员设计更有效的配对系统。了解游戏加速技能获得的哪些方面也提供了一个机会，以设计更有效的教程，设计反馈系统，通过建议游戏行为的改变来帮助新手和专业玩家最大化他们的学习速度，并优化竞争中的训练行为。除了游戏，技能获得研究对于外科和军事航空等领域非常重要，在这些领域，技术进步和性能条件往往超过熟练人员适应它们的速度。了解如何优化学习可以对这些领域产生重大影响，因为手术的成功可能取决于个人适时适应的能力。以前对专业知识和技能获得的研究产生了许多关于练习、反馈和迁移的知识，但主要依赖于实验室结果和对练习行为的回顾描述。通过对来自电子竞技的大型真实世界数据集应用统计建模和数据可视化技术，目前的研究将绕过实验室方法的缺陷，在高统计置信度下生成关于技能学习的原创结果。综上所述，拟议研究的影响可以为电子竞技中的教练实践和游戏开发提供信息，改善我们目前对支撑技能获得的心理学的理解，并激励后续对学习领域的科学探索。关键词词汇预测技能预测机器学习运动分析性能分析

Expert knowledge and skill play an increasingly important role in society, with an economy which is increasingly knowledge-driven and which involves rapid technological change and longer and longer working lives. Game players develop profound skill over hours, days and even years of practice, generating huge reservoirs of data about gameplay and skill development that can be easily and unobtrusively recorded. The result is an unparalleled opportunity to investigate what aspects of play determine expert skill, and how it is acquired.The topic of the proposed PhD is to test theories of skill acquisition and characterise expert skill in gaming through the statistical interrogation of naturally occurring data in eSports games. The research would be conducted under the joint supervision of Professor Anders Drachen from the University of York and Dr Tom Stafford from the University of Sheffield under the IGGI research theme of using eSports data to understand human psychology. The proposed research fits well with available links to the eSports industry (i.e., DC Labs), available expertise (e.g., Professor Alex Wade), and ongoing analytics at IGGI into performance and skill acquisition through the analysis of player data from major online games (e.g., League of Legends, DOTA 2, Destiny)The cognitive science of skill acquisition has broad applicability and is of practical significance for the eSports industry. Developing measures of skill that are predictive of high level performance can allow professional teams to identify talent more effectively, and allow game developers to engineer more effective matchmaking systems. Understanding what aspects of play accelerate skill acquisition also presents an opportunity to design more effective tutorials, engineer feedback systems that help novice to professional players maximise their rate of learning by suggesting changes to gameplay behaviour, and optimise training behaviours for competition.Beyond gaming, skill acquisition research is of tremendous import for domains such as surgery and military aviation, where technical advancements and performance conditions often outpace the rate at which skilled personnel can adapt to them. Understanding how to optimise learning can significantly impact these arenas, as the success of an operation can depend upon an individual's ability to adapt in good time.Previous research into expertise and skill acquisition has generated much knowledge about practice, feedback, and transfer, but has largely relied on laboratory findings and retrospective accounts of practice behaviour. By applying statistical modelling and data visualisation techniques on large, real-world data sets from eSports, the current research would bypass the drawbacks of laboratory methods to generate original findings about skill learning at high levels of statistical confidence. Taken together, the impact of the proposed research could inform coaching practices and game development in eSports, improve our current understanding about the psychology underpinning skill acquisition, and motivate subsequent scientific inquiry into the domain of learning.Keywordsprediction modelingskill predictionmachine learningsports analyticsperformance analysis