Kepler.gl Dataset Format

Kepler.gl Dataset Format#

Native Kepler.gl dataset configuration format reference.

When using KeplerDataset(raw_dict={...}), the dictionary should follow the native Kepler.gl dataset structure documented here.

Dataset Structure#

A Kepler.gl dataset configuration has the following structure:

{
    "version": "v1",
    "data": {
        "id": "dataset-id",          # Dataset identifier
        "label": "Dataset Label",    # Display label
        "color": [255, 0, 0],        # Optional RGB color
        "allData": [...],            # Array of data rows (optional)
        "fields": [...]              # Array of field definitions (optional)
    }
}

Fields Definition#

Each field in the fields array describes a column:

{
    "name": "column_name",        # Column name
    "type": "string",             # Data type: string, integer, real, boolean, timestamp, geometry
    "format": "",                 # Optional format string
    "analyzerType": "STRING"      # Analyzer type: STRING, INT, FLOAT, BOOLEAN, DATE, GEOMETRY
}

Common Field Types#

  • string / STRING: Text data

  • integer / INT: Integer numbers

  • real / FLOAT: Floating point numbers

  • boolean / BOOLEAN: True/false values

  • timestamp / DATE: Temporal data

  • geometry / GEOMETRY: Spatial data (GeoJSON)

Examples#

Basic Dataset#

from graphistry import KeplerDataset

dataset = KeplerDataset({
    "version": "v1",
    "data": {
        "id": "cities",
        "label": "City Locations",
        "color": [255, 140, 0]
    }
})

Dataset with Fields#

dataset = KeplerDataset({
    "version": "v1",
    "data": {
        "id": "points",
        "label": "Points of Interest",
        "fields": [
            {"name": "name", "type": "string", "analyzerType": "STRING"},
            {"name": "latitude", "type": "real", "analyzerType": "FLOAT"},
            {"name": "longitude", "type": "real", "analyzerType": "FLOAT"},
            {"name": "count", "type": "integer", "analyzerType": "INT"},
            {"name": "timestamp", "type": "timestamp", "analyzerType": "DATE"}
        ]
    }
})

See Also#