Meaning
Bayesian statistical range representing the narrowest region on a posterior distribution that contains a fixed percentage of total probability. A highest density interval is used to provide a summary of an estimate that is more precise than a central interval. It ensures that every value inside the range has a higher probability density than any value outside of it.
This makes it an efficient way to describe the likely performance of a manufacturing process.
Geometric Efficiency
Coverage of the probability is achieved with the minimum possible distance between the upper and lower bounds. This property is useful when the underlying data distribution is skewed or has multiple peaks.
Decision Boundary
Using this range allows a quality engineer to set tighter tolerances for a production run. It focuses attention on the most likely outcomes rather than the theoretical extremes of a wider confidence band.
Process Stability
Frequent shifts in the location of this interval indicate an unstable process that requires calibration. Steady intervals suggest that the system is operating within its designed capability. Calculating this interval requires significant computational power for complex models.
The result is a defensible measure of the readiness of a prototype for mass production.