Vega Datasets

0.9.0 · deprecated · verified Thu Apr 16

A Python package providing convenient access to a collection of over 70 datasets used in Vega, Vega-Lite, and Altair examples and documentation. As of Altair 6.0.0, the `vega-datasets` package has been archived, and its functionality, including all datasets, has been integrated directly into the `altair.datasets` module. The last standalone release of `vega-datasets` was 0.9.0.

Common errors

```
ModuleNotFoundError: No module named 'vega_datasets'
```
cause The `vega_datasets` package is not installed, or you are attempting to use the old import path (`from vega_datasets import data`) after migrating to Altair 6.0.0+ without installing the standalone `vega_datasets` package.

fix If working with older Altair versions, ensure `vega-datasets` is installed (`pip install vega-datasets`). If using Altair 6.0.0+, switch your import to `from altair.datasets import data` and ensure `altair` is installed (`pip install altair`).
```
MaxRowsError: The number of rows in your dataset is greater than the maximum allowed (5000).
```
cause Altair's default data transformer limits the size of embedded datasets to prevent performance issues in visualizations.

fix To process larger datasets, enable the VegaFusion data transformer: `import altair as alt; alt.data_transformers.enable('vegafusion')`. This requires `vegafusion` to be installed (`pip install vegafusion`). For extremely large datasets, consider passing data by URL.
```
Data not loading, chart appears empty or incomplete in browser or Jupyter notebook.
```
cause Browser ad-blockers, privacy extensions, or strict Content Security Policies (CSPs) might block requests to CDNs (e.g., `cdn.jsdelivr.net`) where some `vega-datasets` are hosted.

fix Temporarily disable ad-blockers or privacy extensions for the specific page/domain. Check your browser's developer console (Network tab) for blocked requests or `ERR_BLOCKED_BY_CLIENT` messages. Consider using datasets bundled with Altair or serving them locally.

Warnings

breaking The `vega-datasets` package has been archived and its functionality migrated to `altair.datasets` in Altair 6.0.0+. Direct `import vega_datasets` will no longer receive updates and is discouraged for new projects.
Fix: Migrate `from vega_datasets import data` to `from altair.datasets import data` and ensure Altair 6.0.0 or higher is installed (`pip install altair>=6.0.0`).
deprecated The standalone `vega-datasets` PyPI package (version 0.9.0 and earlier) is deprecated. While it remains installable, new datasets and updates are only provided via `altair.datasets`.
Fix: Update to Altair 6.0.0+ and use `altair.datasets` for consistent access to all Vega datasets.
gotcha Dataset names containing hyphens (e.g., 'sf-temps') must be accessed using underscores (e.g., `data.sf_temps()`) when using the Python interface.
Fix: Replace hyphens with underscores when calling dataset methods (e.g., `data.dataset_name()`).
gotcha When visualizing large datasets with Altair, you might encounter a `MaxRowsError`. This is a default safeguard in Altair to encourage efficient data handling.
Fix: Enable the VegaFusion data transformer (`import altair as alt; alt.data_transformers.enable('vegafusion')`) by installing `vegafusion` (`pip install vegafusion`). Alternatively, pass data by URL if the dataset is hosted remotely, possibly using `altair_data_server` for local files.

Install

pip install vega-datasets Install standalone package (deprecated)
pip install altair>=6.0.0 Install Altair with built-in datasets (recommended)

Imports

data
wrong:
```
from vega_datasets import data
```
correct:
```
from altair.datasets import data
```
The `vega_datasets` package is archived. For Altair versions 6.0.0 and above, datasets are accessed via `altair.datasets`.

Quickstart

Demonstrates loading the 'cars' dataset using both the deprecated `vega_datasets` import and the recommended `altair.datasets` import for newer Altair versions. The datasets are loaded as Pandas DataFrames.

# Old way (vega-datasets 0.9.0 and earlier)
# Requires: pip install vega-datasets
try:
    from vega_datasets import data
    cars_df_old = data.cars()
    print("Old import (vega-datasets 0.9.0):\n", cars_df_old.head())
    # Accessing metadata
    # print(data.cars.description)
except ImportError:
    print("vega-datasets not installed, skipping old import example.")

# New way (Altair 6.0.0+ with built-in datasets)
# Requires: pip install altair>=6.0.0
try:
    from altair.datasets import data as altair_data
    cars_df_new = altair_data.cars()
    print("\nNew import (Altair 6.0.0+):\n", cars_df_new.head())
    # Accessing metadata
    # print(altair_data.cars.description)
except ImportError:
    print("\nAltair 6.0.0+ not installed or older version. Cannot use new import example.")

view raw JSON →