2022LoLProAnalysis

Exploring how Side Selection and Early Game Statistics Affect Professional Match Results

Overview

This project analyzes a 2022 League of Legends Esports dataset to explore how side selection and early game statistics affect match outcomes. It aims to uncover trends and insights that can drive strategic decision-making for professional teams.

Dataset Description

• Side Selection: Indicates which side (red or blue) a team plays on.
• Early Game Statistics:
  • Combined metrics: Gold, Experience, and Creep Score at both the 10 and 15 minute marks.
• Match Outcome:
  • Result: Encoded as 1 (win) or 0 (loss), where the mean value is directly proportional to the win rate.

Key Findings and Insights

• Blue Side Advantage:

  • Teams playing on the blue side exhibit a higher overall win rate compared to the red side.

• Statistical Differences:

  • At both the 10 and 15 minute marks, blue side teams show a positive mean difference in combined statistics, whereas red side teams show a negative difference.
  • Blue side teams also have a higher KDA (kill-death-assist) ratio compared to red side teams.

• Correlation with Win Rate:

  • Differences in gold, experience, and overall statistics are directly proportional to win rates, indicating that early game performance is a strong predictor of match outcomes.

Leveraging the Dataset for Decision-Making

In professional League of Legends, especially in best-of formats (best-of-3 or best-of-5), a team that loses a game gains priority in side selection for subsequent games. Given the statistically higher win rate for blue side teams, this analysis suggests that teams can optimize their strategy by:

• Prioritizing blue side selection when given the choice.
• Using early game metrics to predict match outcomes and adjust tactics accordingly.

Suitable Machine Learning Problem Type

Based on the analysis, a classification problem is most appropriate for this dataset:

• Target Variable:

  • The match outcome (win=1, loss=0) makes it a binary classification problem.

• Rationale:

  • The goal is to predict the probability of a win based on early game statistics and side selection.
  • Common classification algorithms like logistic regression, decision trees, random forests, or SVM can be applied to model the relationship between the predictors and the match result.

• Application:

  • A trained classification model can serve as a decision support tool for professional teams, informing them about the potential impact of side selection and early game performance on overall match outcomes.

Conclusion

This analysis demonstrates that early game metrics and side selection play a crucial role in determining match outcomes in professional League of Legends. By applying classification techniques, teams can leverage historical data to make informed decisions, potentially gaining a competitive edge in tournaments.