diff --git a/content/posts/debiased_rankings/presentation.html b/content/posts/debiased_rankings/presentation.html new file mode 100644 index 0000000000000000000000000000000000000000..1bf03ce4d342dc6f6d834458760e9923f98a7e82 --- /dev/null +++ b/content/posts/debiased_rankings/presentation.html @@ -0,0 +1,190 @@ +
+

Debiasing the BoardGameGeek (BGG) Ranking

+

Understanding and Adjusting for Biases in Game Ratings

+
Debiasing BGG Rankings | Recommend.Games
+
+
+

What is Bias?

+
    +
  • Bias in statistics: A systematic deviation in data.
    • +
  • BoardGameGeek Biases: +
      +
    • Age Bias: Newer games tend to be rated higher.
      • +
    • Complexity Bias: Heavier games get higher ratings.
      • +
    • Playing Time Bias: Longer games receive higher scores.
      • +
    • Game Type Bias: Some categories are favoured over others.
      • +
    • Co-op Bias: Cooperative games tend to score higher.
      • +
    +
    • +
+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Age Bias in BGG Rankings

+

📊 Observation: Newer games receive higher ratings.

+

🔍 Data Insight:

+
    +
  • On average, a game's rating drops by 0.03 per year since release.
    • +
+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Age Bias - Visualization

+

🖼️ Scatter Plot

+

Age Bias Animation

+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Adjusting for Age Bias

+

✅ Method: Normalize ratings to remove age-based inflation.

+

🎯 Impact: Older games regain prominence.

+

🏆 Revised Top 3 Games:

+
    +
  1. Gloomhaven (2017): 8.4
    2. +
  2. Twilight Struggle (2005): 8.4
    3. +
  3. Pandemic Legacy: Season 1 (2015): 8.4
    4. +
+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Complexity Bias in BGG Rankings

+

📊 Observation: Heavier games are often rated higher.

+

🔍 Data Insight:

+
    +
  • The heaviest games receive, on average, 2.5 points more than the lightest.
    • +
+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Complexity Bias - Visualization

+

🖼️ Scatter Plot

+

Complexity Bias Animation

+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Adjusting for Complexity Bias

+

✅ Method: Normalize ratings based on complexity.

+

🎯 Impact: Lighter games receive a fairer ranking.

+

🏆 Revised Top 3 Games:

+
    +
  1. Just One (2018): 8.2
    2. +
  2. Codenames (2015): 8.0
    3. +
  3. Cascadia (2021): 8.0
    4. +
+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Playing Time Bias in BGG Rankings

+

📊 Observation: Longer games tend to receive higher ratings.

+

🔍 Data Insight:

+
    +
  • A game's rating increases by 0.0045 per minute of playing time.
    • +
+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Playing Time Bias - Visualization

+

🖼️ Scatter Plot

+

Playing Time Bias Animation

+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Adjusting for Playing Time Bias

+

✅ Method: Normalize ratings based on game length.

+

🎯 Impact: Shorter games receive adjusted scores.

+

🏆 Revised Top 3 Games:

+
    +
  1. Game A: Adjusted Rating
    2. +
  2. Game B: Adjusted Rating
    3. +
  3. Game C: Adjusted Rating
    4. +
+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Game Type Bias in BGG Rankings

+

📊 Observation: Different game types are rated unequally.

+

🔍 Data Insight:

+
    +
  • Strategy and war games tend to dominate rankings, while party and family games lag.
    • +
+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Game Type Bias - Visualization

+

🖼️ Category Plot

+

Game Type Bias Animation

+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Adjusting for Game Type Bias

+

✅ Method: Normalize ratings across game types.

+

🎯 Impact: Categories are balanced to reflect game quality fairly.

+

🏆 Revised Top 3 Games:

+
    +
  1. Brass: Birmingham (2018): 8.2
    2. +
  2. Star Wars: Rebellion (2016): 8.2
    3. +
  3. Ark Nova (2021): 8.1
    4. +
+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Cooperative Game Bias in BGG Rankings

+

📊 Observation: Cooperative games score higher than competitive ones.

+

🔍 Data Insight:

+
    +
  • Co-op games have an average rating 0.48 points higher than competitive ones.
    • +
+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Cooperative Game Bias - Visualization

+

🖼️ Category Plot

+

Co-op Bias Animation

+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Adjusting for All Biases

+

✅ Method: Apply a multivariate model to adjust for all factors at once.

+

🎯 Impact: The ultimate debiased ranking removes age, complexity, playing time, game type, and co-op bias simultaneously.

+

🏆 Revised Top 3 Games:

+
    +
  1. Survive: Escape from Atlantis! (1982): 8.2
    2. +
  2. Can't Stop (1980): 8.2
    3. +
  3. For Sale (1997): 7.9
    4. +
+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Conclusion

+

📢 Summary:

+
    +
  • BGG rankings contain multiple biases that affect game visibility.
    • +
  • Debiasing reveals hidden classics and rebalances rankings.
    • +
+

🚀 Implication:

+
    +
  • Players can discover great games without being influenced by trends.
    • +
  • A more accurate and fair ranking benefits the community.
    • +
+

🎲 Play more, discover better!

+
Debiasing BGG Rankings | Recommend.Games
+
+
+

Thank You! 🎉

+

Questions?

+

🔗 blog.recommend.games

+
Debiasing BGG Rankings | Recommend.Games
+
+
\ No newline at end of file diff --git a/content/posts/debiased_rankings/presentation.md b/content/posts/debiased_rankings/presentation.md new file mode 100644 index 0000000000000000000000000000000000000000..9e70e76a74fe51e06531fba587d1a58dfcf02722 --- /dev/null +++ b/content/posts/debiased_rankings/presentation.md @@ -0,0 +1,174 @@ +--- +marp: true +theme: default +paginate: true +footer: "Debiasing BGG Rankings | Recommend.Games" +--- + +# Debiasing the BoardGameGeek (BGG) Ranking +### Understanding and Adjusting for Biases in Game Ratings + +--- + +## What is Bias? +- **Bias in statistics**: A systematic deviation in data. +- **BoardGameGeek Biases**: + - **Age Bias**: Newer games tend to be rated higher. + - **Complexity Bias**: Heavier games get higher ratings. + - **Playing Time Bias**: Longer games receive higher scores. + - **Game Type Bias**: Some categories are favoured over others. + - **Co-op Bias**: Cooperative games tend to score higher. + +--- + +## Age Bias in BGG Rankings +📊 **Observation**: Newer games receive higher ratings. + +🔍 **Data Insight**: +- On average, a game's rating **drops by 0.03** per year since release. + +--- + +## Age Bias - Visualization +🖼️ **Scatter Plot** + +![Age Bias Animation](plot_reg_age_animated.gif) + +--- + +## Adjusting for Age Bias +✅ **Method**: Normalize ratings to remove age-based inflation. + +🎯 **Impact**: Older games regain prominence. + +🏆 **Revised Top 3 Games**: +1. **Gloomhaven** (2017): 8.4 +2. **Twilight Struggle** (2005): 8.4 +3. **Pandemic Legacy: Season 1** (2015): 8.4 + +--- + +## Complexity Bias in BGG Rankings +📊 **Observation**: Heavier games are often rated higher. + +🔍 **Data Insight**: +- The heaviest games receive, on average, **2.5 points more** than the lightest. + +--- + +## Complexity Bias - Visualization +🖼️ **Scatter Plot** + +![Complexity Bias Animation](plot_reg_complexity_animated.gif) + +--- + +## Adjusting for Complexity Bias +✅ **Method**: Normalize ratings based on complexity. + +🎯 **Impact**: Lighter games receive a fairer ranking. + +🏆 **Revised Top 3 Games**: +1. **Just One** (2018): 8.2 +2. **Codenames** (2015): 8.0 +3. **Cascadia** (2021): 8.0 + +--- + +## Playing Time Bias in BGG Rankings +📊 **Observation**: Longer games tend to receive higher ratings. + +🔍 **Data Insight**: +- A game's rating **increases by 0.0045 per minute of playing time**. + +--- + +## Playing Time Bias - Visualization +🖼️ **Scatter Plot** + +![Playing Time Bias Animation](plot_reg_min_time_animated.gif) + +--- + +## Adjusting for Playing Time Bias +✅ **Method**: Normalize ratings based on game length. + +🎯 **Impact**: Shorter games receive adjusted scores. + +🏆 **Revised Top 3 Games**: +1. **Game A**: Adjusted Rating +2. **Game B**: Adjusted Rating +3. **Game C**: Adjusted Rating + +--- + +## Game Type Bias in BGG Rankings +📊 **Observation**: Different game types are rated unequally. + +🔍 **Data Insight**: +- Strategy and war games tend to dominate rankings, while party and family games lag. + +--- + +## Game Type Bias - Visualization +🖼️ **Category Plot** + +![Game Type Bias Animation](plot_cat_game_type_animated.gif) + +--- + +## Adjusting for Game Type Bias +✅ **Method**: Normalize ratings across game types. + +🎯 **Impact**: Categories are balanced to reflect game quality fairly. + +🏆 **Revised Top 3 Games**: +1. **Brass: Birmingham** (2018): 8.2 +2. **Star Wars: Rebellion** (2016): 8.2 +3. **Ark Nova** (2021): 8.1 + +--- + +## Cooperative Game Bias in BGG Rankings +📊 **Observation**: Cooperative games score higher than competitive ones. + +🔍 **Data Insight**: +- Co-op games have **an average rating 0.48 points higher** than competitive ones. + +--- + +## Cooperative Game Bias - Visualization +🖼️ **Category Plot** + +![Co-op Bias Animation](plot_cat_cooperative_animated.gif) + +--- + +## Adjusting for All Biases +✅ **Method**: Apply a multivariate model to adjust for all factors at once. + +🎯 **Impact**: The **ultimate debiased ranking** removes age, complexity, playing time, game type, and co-op bias simultaneously. + +🏆 **Revised Top 3 Games**: +1. **Survive: Escape from Atlantis!** (1982): 8.2 +2. **Can't Stop** (1980): 8.2 +3. **For Sale** (1997): 7.9 + +--- + +## Conclusion +📢 **Summary**: +- BGG rankings contain **multiple biases** that affect game visibility. +- **Debiasing reveals hidden classics** and rebalances rankings. + +🚀 **Implication**: +- Players can discover great games without being influenced by trends. +- A **more accurate and fair ranking** benefits the community. + +🎲 **Play more, discover better!** + +--- + +## Thank You! 🎉 +### Questions? +🔗 [blog.recommend.games](https://blog.recommend.games/posts/debiasing-boardgamegeek-ranking/)