Monte carlo estimation for guaranteed-coverage non-normal tolerance intervals

Abstract

We propose a Monte Carlo sampling algorithm for estimating guananteed-coverage tolerance factors for non-normal continuous distributions with known shape but u n p w n location and scale. The algorithm is based on reformulating this root-finding problem as a quantile-estimation problem. The reformulation leads to a geometrical interpretation of the tolerance-interval factor. For arbitrary distribution shapes, we analytically and empirically investigate various relationships among tolerance- interval coverage, confidence, and sample size.

Keywords:

• Quantile
• Reliability
• Root Finding
• Stochastic Approximation