Update docs for vega-datasets>=2.0.0

[eitanlees]

Important

This issue originally targeted vega-datasets@v2.0.0 via altair-viz/vega_datasets.
Many of the changes are still relevant, but we will be targeting vega-datasets@v3.0.0 by using #3631

Major releases from vega/vega-datasets have the potential to break examples throughout the altair docs.
As a result many examples will need to be updated

Tracking

• 3.0.0 Release vega-datasets#654
• feat(RFC): Adds altair.datasets #3631

Changes

From the Example Gallery

Broken

• Binned Scatterplot due to movies
• Interactive Chart with Cross-Highlight due to movies
• Seattle Weather Heatmap due to seattle_temps
• Top K Items due to movies
• Top-K plot with Others due to movies
• Binned Heatmap due to movies

Needs Update

• Faceted Density Estimates due to iris
• Stacked Density Estimates due to iris
• Multifeature Scatter Plot due to iris

From the docs

Broken

• Bin transforms due to movies
• Density Transform due to movies
• Join Aggregate Transform due to movies
• TimeUnit Within Encoding due to seattle_temps
• TimeUnit as a Transform due to seattle_temps
• Window Transform due to movies
• Order of Layers due to movies
• Horizontal Concatenation due to iris
• Repeated Charts due to iris
• Faceted Charts due to iris

Needs Update

• Adjusting the Legend due to iris
• Removing the Chart Border due to iris
• Color Schemes due to iris
• Color Domain and Range due to iris
• Altair and Pandas Datetimes due to seattle_temps
• Specifying Time Zones due to seattle_temps
• Using UTC Time due to seattle_temps

[jakevdp]

Related: #2210, #2212

[jakevdp]

Note that some docs are currently broken due to the changes in vega/vega-datasets, due to using URL data that has changed or been removed in vega-datasets.

[eitanlees]

I have compiled a big list of everything that needs to be changed once altair-viz/vega_datasets#40 is merged.

Moved to issue description

From the Example Gallery

Broken

• Binned Scatterplot due to movies
• Interactive Chart with Cross-Highlight due to movies
• Seattle Weather Heatmap due to seattle_temps
• Top K Items due to movies
• Top-K plot with Others due to movies
• Binned Heatmap due to movies

Needs Update

• Faceted Density Estimates due to iris
• Stacked Density Estimates due to iris
• Multifeature Scatter Plot due to iris

From the docs

Broken

• Bin transforms due to movies
• Density Transform due to movies
• Join Aggregate Transform due to movies
• TimeUnit Within Encoding due to seattle_temps
• TimeUnit as a Transform due to seattle_temps
• Window Transform due to movies
• Order of Layers due to movies
• Horizontal Concatenation due to iris
• Repeated Charts due to iris
• Faceted Charts due to iris

Needs Update

• Adjusting the Legend due to iris
• Removing the Chart Border due to iris
• Color Schemes due to iris
• Color Domain and Range due to iris
• Altair and Pandas Datetimes due to seattle_temps
• Specifying Time Zones due to seattle_temps
• Using UTC Time due to seattle_temps

[joelostblom]

@eitanlees @jakevdp Is there anything I can do to assist with this that is not already covered by #2231, #2241, and #2310 (I just opened this for the weather data)? It seems like altair-viz/vega_datasets#40 is ready also. I am teaching a course with Altair next week and want to make sure students have a good experience reading the documentation, so let me know if I can help out.

[joelostblom]

Response to my question above here #2310 (comment)

[cdeil]

@jakevdp @ellisonbg - any chance to get the example dataset URL issues resolved for Altair?

I noticed the broken example in the docs and found this issue. Looks like there's open PRs with fixes already.

[jakevdp]

Current broken docs fixed by using vega_datasets 0.9, which loads datasets from CDN to get around the broken URLs introduced by vega/vega-datasets version 2.0.

Going forward, we need to do some more work, mainly to remove the iris examples throughout the docs, so that the docs will be compatible with the datasets now available in vega-datasets 2.0.

The renamed fields in vega-datasets 2.0 are going to be more difficult, because there's not really any way to make Altair's docs compatible with both the old and the new version at once. So we'll need a vega_datasets release simultaneous with an altair release.

[dangotbanned]

Removed the bug label as broken examples were resolved indirectly by altair-viz/vega_datasets#45

The related upstream PR that would have fixed this directly by updating to 2.0.0 was never reviewed

• Update to vega datasets v2 altair-viz/vega_datasets#40

I've updated the description to more accurately reflect current plans