---
license: mit
task_categories:
- tabular-regression
- time-series-forecasting
tags:
- aviation
- adsb
- aircraft-tracking
- trajectories
- reinforcement-learning
- tabular-data
- time-series
size_categories:
- 10K<n<100K
---

# TartanAviation ADS-B Dataset (19.7K Clean Samples)

## Dataset Description

**19,714 high-quality ADS-B trajectory datapoints** from aircraft operations, rigorously cleaned and validated. Perfect for machine learning research in aviation, reinforcement learning, and trajectory prediction.

## Key Features
- 19,714 clean samples (no missing data, no duplicates)  
- 17 comprehensive features including aircraft ID, timestamp components, altitude, speed, heading, geolocation, and metadata  
- Data period: January to October 2022  
- Geographic coverage: Ohio airspace (KBTP airport region)  
- Formats available: CSV (human-readable) and JSONL (ML-optimized)

## Dataset Structure (17 columns)
- Temporal: year, month, day, hour, minute, second  
- Aircraft State: aircraft_id, altitude_ft, ground_speed_kts, heading_deg  
- Location: latitude, longitude  
- Metadata: aircraft_tail, data_age_sec, range_nm, bearing_deg, altitude_is_gnss

## Use Cases
- Reinforcement learning for air traffic management  
- Aircraft trajectory prediction  
- Graph neural networks modeling  
- Aviation safety analysis  
- Time series forecasting for flight dynamics  

## Quick Start

import pandas as pd

Load CSV dataset
df = pd.read_csv('tartanaviation_adsb_19k_clean.csv')
print(f"Dataset shape: {df.shape}")
print(f"Unique aircraft: {df['aircraft_id'].nunique()}")

text

## Dataset Stats
- Samples: 19,714  
- Features: 17  
- Unique Aircraft: 4,946  
- Time Period: January - October 2022  
- Region: Ohio airspace (KBTP airport area)  
- Data Quality: 100% complete, validated

## Citation

@dataset{tartanaviation_adsb_2024,
title={TartanAviation ADS-B Dataset (19.7K Clean Samples)},
author={Pathange},
year={2024},
publisher={Hugging Face},
url={https://huggingface.co/datasets/Pathange/tartanaviation-adsb-19k-clean}
}

text
undefined