What it is
The first is systems a regulator records as being under an annual filing requirement that have filed nothing since. The requirement is published per system, the filings are published per year, and the absence is only visible if you hold both.
The second is systems whose reported figure moves more year to year than most of their state’s does. Real losses do not change that fast. A series that swings is describing the measurement rather than the water, and a figure from it should not be carried into a model.
What you hand us
A portfolio, a program boundary, or the criteria that define the set. If you want the whole of a state, say so — that is the easier request.
What you get back
One row per system, with the requirement it is recorded under, the last year it filed, how many years are missing, and where the second screen applies, how far its series moves against its state’s own distribution. Each row names the file behind it.
What it is built from
Each regulator’s published filings and, where the regulator publishes one, the requirement recorded on the system’s own most recent filing. See what we assemble, and where it comes from.
What it does not do
Absence from the record is not proof a utility failed to file. A filing can be missing from a published export because it was refused at our end for an identifier we could not resolve, and where that has happened the system’s page says so and names the year. The screen reports what the published record contains.
It does not say a lapsed filer is in trouble. Requirements change, small systems move between cycles, and a system can be exempt for reasons the export does not carry.
It does not flag a figure as wrong. An unstable series is one that cannot be relied on for a trend; the individual figures may each be exactly what the utility measured.