Lanternko

Lanternko / ARAM-Mayhem-Database

Public

查詢 大亂鬥 的英雄和增幅裝置！

lanternko.github.ioARAM-Mayhem-Database
17
2
80% credibility
Found May 17, 2026 at 17 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

This project is a machine learning system for League of Legends ARAM Mayhem players. It automatically collects game data from your matches and other players, trains AI models to predict team win rates, and provides a real-time GUI that suggests the best champion swaps during champion selection. The tool analyzes champion synergies and individual strength to help you make smarter decisions when deciding whether to use your reroll.

How It Works

1
🎮 You queue up for ARAM Mayhem

You're excited to play your favorite chaotic game mode and want every advantage you can get.

2
🔗 You connect the tool to League

The tool links to your League client automatically and starts watching your games.

3
📊 Your games are collected and analyzed

Every ARAM Mayhem match you play gets saved, along with thousands of games from other players to teach the AI.

4
🧠 The AI learns champion strengths

The system studies which champions win more often together, building a map of team synergies and individual power levels.

5
During champion selection
✨
Swap recommendation appears

The GUI shows you a ranked list of bench swaps with clear win-probability changes for each option.

📈
Synergy analysis available

You can also see detailed breakdowns of how each champion works with your specific team composition.

🏆 You lock in the best pick

With clear data showing you should swap to that champion, you make your choice confidently and queue into the game.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 17 to 17 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is ARAM-Mayhem-Database?

This is a League of Legends ARAM Mayhem prediction engine built in Python. It collects match data directly from the League client, trains machine learning models to predict champion win rates, and serves real-time swap recommendations during champion select. The system uses PyTorch for neural network training and polars for data processing, with a Tkinter GUI that displays champion icons and win probability deltas while you queue.

Why is it gaining traction?

The project solves a real problem: ARAM's hidden enemy composition makes optimal champion selection feel like guesswork. This tool gives you statistical backing -- it calculates per-champion strength scores, synergy lifts against your specific team comp, and conditional win rates based on co-occurrence data. The GUI updates live as champ select progresses, showing which bench champion would boost your team's win probability most. There's also a CLI for generating tier lists and running ablation experiments.

Who should use this?

Competitive ARAM players who want data-driven swap decisions. Data scientists interested in applied ML on team-composition prediction. League of Legends modders building custom tools. The project requires a running League client for live recommendations, or you can use the Riot API for historical analysis.

Verdict

At 17 stars with a 0.800000011920929% credibility score, this is a personal research project in active experimentation -- not production-ready software. The code quality is high (sanity tests, frozen snapshots, proper train/val splits), but documentation is minimal and the GUI is functional rather than polished. Worth exploring if you're comfortable with Python and want to experiment with team-composition ML, but don't expect a plug-and-play recommendation engine.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.