Rolling average example

[rolyp]

• Regular test case (without slicing) ❌ linked outputs test case better (assumptions about file locations, etc)
• Linked outputs test case (with just identity as expected output selection)
• First pass over example
• Linked outputs test case (with non-trivial expected output selection) ❌ revisit with better dependencies
• Drop central? Or simple?
• Simplify moving-average library
• Remove LinkedText aspect (served better by IPCC AR6: Table SPM.1 #1118)
• Better solution to newPoints workaround (take floor)
• y coordinate not consumed by original curve
• Less trivial/more meaningful data set

---

My original thought was to visualise some kind of cumulative function of a time series. That should show a given output being related to every prior element in the series. But from the Cross-VESRI event I realised it is common to use line charts to plot a time series alongside a rolling average to visualise trends more clearly. For example, one could plot annual global average temperature and overlay it with a 5-year rolling average to smooth out short-term fluctuations and highlight longer-term trends.

What we could implement is a slightly more self-explanatory version of this where the related outputs would be automatically highlighted. (With #268 we could also show the calculation of the average, but it would still be useful just to show the sliding window.)

Maybe this could be coupled with a selection visualisation convention whereby if adjacent points are selected then the line that connects them also has some selection highlighting.

See also:

• LineChart: highlight line segments between adjacent point selections #1073
• Ralph Keeling’s work on model vs. observation line plots of CO2, combined with 5-year averages

[rolyp]

Had a better idea for this and bumped from Proposed to Planned.

[rolyp]

@JosephBond Taking a look at this this morning. Added a subtask to add a test case once we have things working as best as they can given the current issues with the graph.