Health score & stability
How Open Gauge's 0-100 health score is composed, and what "stability" and "recommended interval" mean.
Every asset with at least two calibrations on a channel gets a computed health score (0–100), visible on the asset's Health tab. With only one calibration, there isn't yet a trend to assess — Open Gauge shows a neutral 100% score, "Stable" stability, and 0 drift rate rather than a misleading result computed from a single point.
Health score
A weighted composite of six inputs, each individually normalized then combined:
| Input | Weight |
|---|---|
| Maximum drift | 30% |
| RMS drift | 20% |
| RMSE | 15% |
| Uncertainty | 10% |
| Hysteresis | 10% |
| Linearity | 10% |
| Trend | 5% |
Maximum and RMS drift dominate the score (50% combined) because they most directly answer the question a health score exists to answer: is this instrument still measuring what it says it measures? See Drift metrics for what each drift term means mathematically.
Stability
A classification (e.g. Stable / Drifting / Unstable) derived from the long-term drift rate — relative to the channel's full-scale span — and the drift trend's regression fit quality. A sensor can have a "good" (high) health score today while still being classified as drifting, if the trend is moving in a worrying direction even though it hasn't crossed a threshold yet.
Average drift
The long-term drift rate from a linear regression over the full calibration history for this channel — how fast, on average, this sensor's error is growing over time (independent of any single calibration's individual max error).
Recommended interval
A heuristic suggestion for the next calibration interval, based on the current stability classification, starting from the sensor's configured default calibration interval. This is an Open Gauge-internal recommendation for planning purposes only — it is never printed on the calibration certificate itself; per ISO/IEC 17025 §7.8.4.3, a certificate only states an interval recommendation when explicitly agreed with the customer, and Open Gauge doesn't have that agreement workflow yet (see Certificate generation).
The charts
- Drift Evolution — maximum drift of each calibration vs. the baseline (first) calibration, with a linear trend line. The primary view of sensor ageing over time.
- Calibration Stability — the historical evolution of RMSE, Maximum Error, Expanded Uncertainty, Hysteresis, and R² across every calibration on this channel, plotted together so you can spot correlated shifts (e.g. uncertainty growing alongside RMSE).
- Calibration Curve Comparison — evaluates the fitted polynomials of two calibrations over roughly 200 points across their shared operating range, isolating how the curve itself has changed shape — independent of any single point's error.