thehooklab/nsw-property

---
license: mit
task_categories:
- tabular-regression
- time-series-forecasting
language:
- en
tags:
- real-estate
- housing
- australia
- property-prices
- nsw
- property-market
- housing-data
size_categories:
- 1M<n<10M
configs:
- config_name: sales
data_files: nsw_property_sales_master.parquet
default: true
- config_name: enriched
data_files: nsw_property_master_enriched.parquet
- config_name: rental
data_files: nsw_rental_market_master.parquet
---

# NSW Property Dataset

**The world's largest open-source dataset on New South Wales property data.**

This comprehensive dataset contains over 3.6 million records spanning property sales, enriched property analytics, and rental market data for New South Wales, Australia. It covers 35 years of historical transactions from 1990 to 2024.

## Dataset Overview

| Config | File | Records | Description |
|--------|------|---------|-------------|
| `sales` (default) | `nsw_property_sales_master.parquet` | 3,493,361 | Historical property transactions (1990-2024) |
| `enriched` | `nsw_property_master_enriched.parquet` | 83,548 | Properties with market analytics and demographics |
| `rental` | `nsw_rental_market_master.parquet` | 48,158 | Rental bond records with pricing data |

**Total: 3,625,067 records**

## Usage

```python
from datasets import load_dataset

# Load sales data (default configuration)
sales = load_dataset("thehooklab/nsw-property")

# Load enriched properties with analytics
enriched = load_dataset("thehooklab/nsw-property", "enriched")

# Load rental market data
rental = load_dataset("thehooklab/nsw-property", "rental")

# Example: Filter Sydney properties over $1M
sydney_premium = sales.filter(lambda x: x['locality'] == 'SYDNEY' and x['purchase_price'] > 1000000)
```

## Data Description

### Sales Data (3.49M records)

Historical property transactions across NSW from 1990 to 2024.

| Column | Type | Description |
|--------|------|-------------|
| `property_id` | float | Unique property identifier |
| `house_number` | string | Street number |
| `street_name` | string | Street name |
| `locality` | string | Suburb/locality name |
| `postcode` | float | Australian postcode |
| `contract_date` | string | Date of sale contract |
| `purchase_price` | float | Sale price in AUD |
| `area` | float | Property area |
| `zone` | string | Zoning classification |
| `property_type` | string | Type (HOUSE, UNIT, VACANT LAND, etc.) |
| `sale_year` | int | Year of sale |
| `lat`, `lon` | float | Geographic coordinates |
| `num_bed`, `num_bath`, `num_parking` | float | Property features |
| `land_area` | float | Land size in sqm |
| `km_from_cbd` | float | Distance from Sydney CBD |
| `suburb_population` | float | Population of suburb |
| `suburb_median_income` | float | Median income of suburb |

### Enriched Properties (83.5K records)

Properties enhanced with market analytics, demographic indicators, and transport accessibility scores.

| Column | Type | Description |
|--------|------|-------------|
| `avg_price` | float | Average historical price |
| `median_price` | float | Suburb median price |
| `price_category` | string | Affordable, Mid-range, Premium, Luxury |
| `market_heat` | string | Market activity indicator |
| `growth_potential` | string | Price growth forecast |
| `rental_yield_estimate` | float | Estimated rental yield % |
| `days_on_market_avg` | float | Average days to sell |
| `auction_clearance_estimate` | float | Auction success rate |
| `family_friendly_score` | float | Family suitability score |
| `walkability_score` | float | Pedestrian accessibility |
| `cafe_culture_score` | float | Local amenities score |
| `school_rating` | string | School quality rating |
| `crime_index` | float | Area crime indicator |
| `gentrification_index` | float | Neighborhood change indicator |
| `nearest_station` | string | Closest train station |
| `distance_to_station_km` | float | Distance to station |
| `transport_score` | float | Public transport accessibility |

### Rental Market (48.2K records)

Rental bond lodgement data with weekly rent and dwelling information.

| Column | Type | Description |
|--------|------|-------------|
| `lodgement_date` | string | Bond lodgement date |
| `postcode` | float | Property postcode |
| `dwelling_type` | string | House, Unit, Townhouse, etc. |
| `bedrooms` | float | Number of bedrooms |
| `weekly_rent` | float | Weekly rent in AUD |
| `bond_amount` | float | Bond amount in AUD |
| `annual_rent` | float | Calculated annual rent |

## Use Cases

- **Property Price Prediction**: Train ML models to predict property values
- **Market Analysis**: Analyze price trends across suburbs and time periods
- **Investment Research**: Identify high-growth areas and rental yields
- **Urban Planning**: Study development patterns and demographic shifts
- **Academic Research**: Housing affordability, market dynamics studies

## Data Quality

- All numeric columns validated and cleaned
- String columns properly typed to avoid parsing errors
- Coordinates available for properties with geolocation data
- Data sourced from official NSW government records and verified sources

## Citation

If you use this dataset in your research or projects, please cite:

```bibtex
@dataset{nsw_property_2026,
title={NSW Property Dataset: The World's Largest Open-Source NSW Property Data},
author={Fedor Kriuk},
year={2026},
publisher={HuggingFace},
url={https://huggingface.co/datasets/thehooklab/nsw-property}
}
```

## License

MIT License - Free to use for commercial and non-commercial purposes.

## Contributing

Found an issue or have suggestions? Open an issue on the dataset repository.