Weighted sum?

It’d be nice to have a weighted sum reducer. This would require two channels as input (the values, and the weights), so I’m not sure how we would specify it. As a workaround, you can write a custom reducer, assuming that your data is indexable:

Plot.ruleX(
  lockdown,
  Plot.groupZ(
    {
      x: {
        reduce(I, X) {
          return (
            d3.sum(I, (i) => lockdown[i].pop * X[i]) /
            d3.sum(I, (i) => lockdown[i].pop)
          );
        }
      }
    },
    { x: "pct_change" }
  )
)

Start with the Plot.groupZ reducer API and the custom reduce workaround shown in the issue, including its use of d3.sum. Define how a weighted sum reducer would receive values and weights, then implement and document the chosen interface; done means weighted aggregation works without the custom reducer.