Characterizing and Quantifying Expert Input Behavior in League of Legends

Title of the Paper

Characterizing and Quantifying Expert Input Behavior in League of Legends

Paper Information

• Subject Area: Analysis of esports players' input behavior and skill training
• Keywords: League of Legends, esports, input skills, player behavior, skill evaluation, visualization training, human-computer interaction, expert skills, game learning, data analysis

Research Background and Issues

• Problems and Challenges:
  • Esports players need to effectively perform tasks using a mouse and keyboard, but comprehensive analyses of input skills are scarce, and data collection is often constrained by environmental limitations.
  • Existing esports data analysis techniques lack granularity, interpretability, and real-time applicability.
  • Although input behavior data has potential for evaluating esports players' performance, the complexity of the data and the lack of standardized behavioral metrics hinder progress in this field.
• Significance:
  • Input skills are a critical foundation for esports players to successfully complete in-game tasks. Analyzing input skills can provide more specific and actionable training recommendations.
  • The rise of esports and its cultural impact have increased the demand and value for in-depth studies of player skills.
• Motivation and Related Work:
  • Previous research has shown that data-driven analysis methods have significant potential in esports training and learning. However, there has been little exploration of low-level input behaviors, with most studies focusing on match outcome statistics or specific game activity logs.
  • For League of Legends, prior research has offered limited understanding of input skills, despite the game being one of the most popular esports titles.

Proposed Solution

• Proposed Method:
  • Develop a comprehensive research framework for input behavior analysis, including:
    1. Designing standardized metrics for expert input behavior.
    2. Collecting and analyzing large-scale input behavior data.
    3. Implementing training interventions through visualization tools.
• Innovations:
  • Introducing specific quantitative metrics for game input skills for the first time.
  • Data collection encompasses a wide range of player skill levels, surpassing the limitations of previous small-scale lab studies.
  • Validating the practical applicability of analysis results through visualization training tools.
• Technical Implementation Steps:
  1. Expert Interviews and Input Skill Index Design:
     • Conducted interviews with five esports players and coaches to identify key input skills and implementation patterns in League of Legends, leading to the design of eight quantitative metrics (e.g., targeting skills, risk compensation skills).
  2. Data Collection and Analysis:
     • Collected input behavior logs from 193 players across 4,835 matches, including detailed keyboard and mouse usage data and match outcomes.
     • Used statistical methods to validate the significant correlation between these input skill metrics and player rankings.
  3. Training Interventions:
     • Developed a visualization tool to help players and coaches understand their input skill levels and train specifically for these skills.
     • Observed changes in player behavior and performance over a three-week trial period.

Research Outcomes

• Specific Results:
  • Designed eight quantitative metrics reflecting the quality of players' input skills, including targeting skills, risk compensation skills, monitoring skills, etc.
  • Analysis results showed that high-ranking players significantly outperformed low-ranking players on these metrics.
  • The dataset was made publicly available as a benchmark for future esports research.
• Advantages Over Existing Solutions:
  • Provides a more granular and interpretable method for analyzing input behavior, surpassing traditional simple metrics like actions per minute (APM).
  • Training software translates analysis results into concrete training recommendations, addressing the lack of standardized tools in esports training.
• Experiment or Evaluation Results:
  • High-ranking players excelled in metrics such as targeting, combo execution, and monitoring skills, demonstrating a deeper understanding of complex game systems.
  • Although the three-week training intervention showed limited statistical significance (p=0.056), it generated positive feedback from players and coaches regarding skill training.
• Limitations and Future Directions:
  • This study focuses on League of Legends; input skills for other esports genres, such as FPS or RTS games, may require redefinition.
  • Lack of standardized player equipment may introduce noise into the research results.
  • The training effects on low-ranking players require further validation.
  • Future research could explore additional behavioral metrics or analyze specific in-game contexts.

This study provides a solid foundation for scientific training in the esports domain and opens new avenues for the analysis and application of input skills.