How Should a Public Tool Explain Estimated Sensor Gaps?
From the earlier reference collection. Read the current software blog.
Show an estimated segment as an estimate and preserve the original missing interval so users can distinguish measurements from reconstruction.
Keep missingness visible
A gap means the source did not provide a measurement for a period. It may reflect a device outage, a collection problem, or a missing record; the gap alone does not establish which cause occurred. Filling it with a line or value can make a chart look continuous, but the display should not suggest that the sensor measured those conditions.
Preserve the original timestamps and missing interval even when a tool offers an estimate. Do not overwrite an absent observation with a generated value in the underlying record. Keeping the raw series separate makes it possible to inspect what was actually received.
Label the method and scope
If a tool estimates values, explain the method in accessible terms and state the interval it covers. A straight line between two measured points, for instance, represents a particular interpolation choice; it does not reveal the true path the variable followed between observations. A flat carry-forward estimate makes a different assumption.
Mark estimated segments in the chart and in downloaded data. A distinct line style, shaded band, or explicit status field can help users tell reconstructed values from direct measurements. Include the endpoints or inputs used, and avoid presenting excessive decimal precision that the method cannot justify.
Use a concrete example
Imagine readings at 09:00 and 11:00 with no observation at 10:00. A tool that displays a value at 10:00 by taking the midpoint between the endpoints should label it as an estimate and identify the interpolation rule. If the measurements differ sharply, that midpoint may not resemble what occurred during the missing hour.
A user examining a peak or threshold crossing could reach a different conclusion depending on whether the gap was measured or filled. Showing the gap and estimated segment separately allows that user to decide whether the reconstructed value is suitable for the task.
Explain what remains unknown
State when estimates are unavailable, how long a gap may be before estimation stops, and whether estimates are excluded from summaries by default. If an estimate is used in an aggregate, disclose that choice and provide a way to distinguish the estimate’s contribution where practical.
A clear public tool treats missingness as information. It may offer a useful reconstruction, but it keeps that reconstruction visibly separate from the source’s actual observations.