Recovering missing data, interpolation and "like" data
If the meter recorded the readings and simply could not send them, they come back on their own when it reconnects. If the meter never recorded them, they cannot be recovered. In that case Augos can fill the gap, but the filled values are not real readings and Augos will always say so before doing it.
Three different situations
1. Buffered, not lost. The meter was offline but recording. It publishes the backlog when it reconnects. Nothing to do but wait. See How current is my data.
2. Never recorded. The meter was without power, or what it sent was missing a parameter, so the readings never existed. Buffering cannot help. This is the only case where the data is genuinely gone.
3. Recorded but not stored. Occasionally records reach Augos and are not written correctly. A background service repairs these. Where a gap predates that service being introduced, it can be repaired manually on request.
The two ways a gap can be filled
Augos offers two remedies for case 2, and it is worth understanding the difference before you accept one.
| What it is | When it is offered | |
|---|---|---|
| Like data | Values copied from comparable days, so the profile looks the way it plausibly would have | Where the profile follows a consistent pattern |
| Interpolation | Average consumption derived from the point's own history | Where a longer period needs filling |
Neither is actual data. Both are clearly identifiable as filled. Augos will not apply either without asking you first, and you should not use filled periods as evidence in a billing dispute or a measurement and verification exercise.
What to do
- Establish which of the three situations you have. Ask support if you are not sure; they can see the publishing log.
- If the data is genuinely missing, decide whether you want the gap filled. For reporting continuity, yes. For anything that has to stand up to scrutiny, usually no.
- Say which periods matter. Support can double-check named periods rather than assuming.
Good to know
- Exports carry an interpolate flag so filled values can be told apart from real ones. If you are doing measurement and verification work, filter on it. Note that a run of zeroes carrying that flag means "not on range" rather than genuine zero consumption, which will otherwise skew a baseline badly.
- When a configuration fault is corrected, ask explicitly whether historical data was affected or only the recent period. The answer is often "only since the issue", but it is worth having in writing.
Still stuck
Email support@augos.io with the point path and the exact period. Say whether you
need the gap filled or left visible.