The CHARGED (City-scale and Harmonized Dataset for Global Electric Vehicle Charging Demand Analysis) dataset provides comprehensive electric vehicle charging data and related auxiliary information for six major cities worldwide: Los Angeles, São Paulo, Shenzhen, Amsterdam, Johannesburg, and Melbourne.
Each city has two data versions:
- Standard version: Complete dataset with all charging sites
xxx_remove_zeroversion: Filtered dataset with charging sites that have zero charge volume, charge duration, and price removed for cleaner analysis
Each city folder contains 10 CSV files. Due to size limitations, some files larger than 100MB are compressed in csv-larger-than-100MB.rar.
- Description: Hourly charging volume (energy consumption) data
- Format: Time series with hourly granularity
- Index: Timestamps (hourly intervals)
- Columns: Site IDs
- Unit: Kilowatt-hours (kWh)
- Use Case: Primary target variable for demand prediction
- Description: Charging duration data representing time spent charging
- Format: Time series with hourly granularity
- Index: Timestamps (hourly intervals)
- Columns: Site IDs
- Unit: Hours (h)
- Note: Values represent charging duration between consecutive time intervals
- Description: Electricity charging fees (energy cost)
- Format: Time series with hourly granularity
- Index: Timestamps (hourly intervals)
- Columns: Site IDs
- Units by City:
- Los Angeles: USD/kWh
- São Paulo: Brazilian Real (BRL)/kWh
- Shenzhen: Chinese Yuan (CNY)/kWh
- Amsterdam: Dutch Guilder/kWh
- Johannesburg: South African Rand (ZAR)/kWh
- Melbourne: Australian Dollar (AUD)/kWh
- Description: Service fees (additional charges beyond electricity cost)
- Format: Time series with hourly granularity
- Index: Timestamps (hourly intervals)
- Columns: Site IDs
- Note: São Paulo, Johannesburg, and Melbourne have no service fees (uniformly set to 0)
- Units: Same as e_price.csv for each respective city
- Description: Comprehensive information about charging sites
- Key Fields:
| Field | Description | Unit |
|---|---|---|
| site_id | Unique identifier for the charging site | N/A |
| longitude | Geographical longitude of site location | Degrees |
| latitude | Geographical latitude of site location | Degrees |
| charger_num | Number of charging chargers at the site | Count |
| total_duration | Total charging duration recorded at site | Hours |
| total_volume | Total charging volume recorded at site | kWh |
| avg_power | Average charging power per charging record | kW |
| perimeter | Perimeter of DBSCAN cluster shape | Meters (m) |
| area | Area of DBSCAN cluster shape | Square meters (m²) |
- Description: Detailed information about individual charging chargers
- Key Fields:
| Field | Description | Unit |
|---|---|---|
| charger_id | Unique identifier for the charging charger | N/A |
| longitude | Geographical longitude of charger location | Degrees |
| latitude | Geographical latitude of charger location | Degrees |
| site_id | Identifier of the site containing this charger | N/A |
| total_duration | Total charging duration at this charger | Hours |
| total_volume | Total energy delivered at this charger | kWh |
| avg_power | Average charging power per charging record | kW |
- Description: Inter-site distance matrix
- Format: Symmetric matrix
- Index/Columns: Site IDs
- Calculation: Geodesic distance based on ellipsoidal Earth model
- Unit: Kilometers (km)
- Use Case: Spatial analysis and clustering
- Description: Hourly weather data from Visual Crossing API
- Source: Visual Crossing Weather API
- Key Fields:
| Field | Description | Unit |
|---|---|---|
| temp | Temperature | Celsius (°C) |
| feelslike | Apparent temperature (heat index/wind chill) | Celsius (°C) |
| humidity | Relative humidity | Percentage (%) |
| dew | Dew point temperature | Celsius (°C) |
| precip | Liquid precipitation amount | mm |
| snow | Snowfall amount | mm |
| snowdepth | Snow depth on ground | mm |
| preciptype | Precipitation type | Array (rain/snow/freezingrain/ice) |
| windgust | Instantaneous wind speed | km/h |
| windspeed | Sustained wind speed | km/h |
| winddir | Wind direction | Degrees |
| pressure | Sea level atmospheric pressure | hPa |
| visibility | Visibility distance | km |
| cloudcover | Cloud coverage | Percentage (0-100%) |
| solarradiation | Solar radiation power | W/m² |
| solarenergy | Solar energy accumulation | MJ/m² |
| uvindex | UV exposure index | Scale (0-10) |
| conditions | Weather conditions | Categorical (see mapping below) |
Weather Conditions Mapping:
- 0: Clear
- 1: Overcast
- 2: Partially cloudy
- 3: Rain
- 4: Rain, Fog
- 5: Rain, Overcast
- 6: Rain, Partially cloudy
- 7: Snow
- 8: Snow, Fog
- 9: Snow, Partially cloudy
- 10: Snow, Rain
- 11: Snow, Rain, Overcast
- 12: Snow, Rain, Partially cloudy
- Description: Points of Interest (POIs) from OpenStreetMap
- Source: OpenStreetMap
- Categories: Amenities, buildings, offices, shops, leisure facilities, etc.
- Reference: OSM Wiki - Map Features
- Fields:
| Field | Description | Unit |
|---|---|---|
| type | Geographic feature category | N/A |
| longitude | POI longitude | Degrees |
| latitude | POI latitude | Degrees |
- Description: Dataset metadata and summary statistics
- Key Fields:
- Anomaly Detection: Outliers identified and corrected using IQR method
- Zero Sequence Handling: Zero sequences interpolated
- Site Clustering: Geographic clustering using DBSCAN algorithm
- Some cities may have missing data during certain time periods
- Weather data availability depends on API coverage
- POI data completeness varies by city and OSM coverage
- Currency exchange rates not provided for cross-city comparisons
For questions about the dataset, please refer to:
- Main project documentation: README.md
- GitHub Issues: Project Issues
- Contact: guozh29@mail2.sysu.edu.cn