Illustrate a change in methodology

BAAQMD / charting-annual-data

Charting Annual Data: An Inventory Cookbook

0 stars 0 forks source link

Illustrate a change in methodology #2

Closed dholstius closed 3 years ago

dholstius commented 4 years ago

Sometimes there is a qualitative break in the basis for a longitudinal dataset. On the face of things, it might look like newer numbers are still commensurable with older numbers, but for the purpose(s) at hand they're not — or at least not without adjustment.

Here are two examples:

Emission data. The method for estimating a component — be it throughputs, emission factors, controls factors, etc. — may change. We often refer to this as "a change in methodology".
Health data. The diagnostic criteria and associated categories for asthma used to be based on ICD-9, but now they're based on ICD10.

The issue here is simply to add an example (or more than one) of how to appropriately and usefully visualize such a discontinuity.

dholstius commented 4 years ago

A good illustrative example would be VMT data, pre/post CY2000. (@ydu0316 has collated that data, I think, in a recent email thread ... could it be pasted here?)

From CY1990 to CY1999, VMT estimates are based on EMFAC2011.
From CY2000 to CY2030, VMT estimates are based on a later vintage of EMFAC.

This could easily be visualized with something like:

vmt_data %>%
  chart_annual_quantities_by(
    color = vmt_basis)

... assuming that vmt_data contained year, vmt_qty, vmt_unit, and vmt_basis. The last (vmt_basis) would take on values "EMFAC2011" and "EMFAC2017" (or whatever the latter should be).

dholstius commented 4 years ago

New, via https://github.com/BAAQMD/charting-annual-data/commit/3e1d6b86f402b8d0765266fa8f99c6d4da5f9d2d: a brief example, now live at https://baaqmd.github.io/charting-annual-data/11-discontinuities.html. The topic came up during today's E&CEA Section meeting.