inputs/AgentSelect/other/data_statistic/analysis/summary.md

# Dataset Audit: PartII

## Overview

- Questions: 76197
- Agents: 47949
- Observed interactions: 76197
- Matrix size: 76197 × 47949 = 3653569953
- Density: 0.000021
- Sparsity: 0.999979

## Tool Coverage

- Tools defined in the catalog: 0
- Unique tools used by agents: 12099

- Warning: 12099 tools used by agents are missing from the catalog: << All-In-One Crypto Swiss Knife &&Address transaction history>>, << All-In-One Crypto Swiss Knife &&Games>>, << All-In-One Crypto Swiss Knife &&NFTs on address>>, << All-In-One Crypto Swiss Knife &&Net worth history (sparklines)>>, << All-In-One Crypto Swiss Knife &&News>>, << All-In-One Crypto Swiss Knife &&Podcasts>>, << All-In-One Crypto Swiss Knife &&Pools>>, << All-In-One Crypto Swiss Knife &&Projects / Staking/ Other Activities>>, << All-In-One Crypto Swiss Knife &&Protocol>>, << All-In-One Crypto Swiss Knife &&Protocols>>, << All-In-One Crypto Swiss Knife &&Tokens Balances>>, << All-In-One Crypto Swiss Knife &&Top holders>>, << All-In-One Crypto Swiss Knife &&Total value locked>>, << All-In-One Crypto Swiss Knife &&Trending games>>, << Cheap YouTube API &&Channel>>, << Cheap YouTube API &&Comments>>, << Cheap YouTube API &&Playlist>>, << Cheap YouTube API &&Popular>>, << Cheap YouTube API &&Search suggestions>>, << Cheap YouTube API &&Search>> ...

## Tools per Agent

- Min / median / mean / p90 / max: 0 / 3 / 2.8232 / 4 / 411

## Generated Figures

1. Tools per agent, raw distribution: `tools_per_agent_hist_raw.png`

2. Tools per agent, filtered distribution: `tools_per_agent_hist_filtered.png` (lower=0.00, upper=10.00, removed=46)

3. Tool popularity, top 20: `tool_popularity_topN.png`

4. Interaction matrix heatmap: skipped to avoid excessive memory and runtime.