hesela.devBinomial proportion
A binomial proportion is an outcome count divided by a fixed number of independent binary trials, used to estimate their common outcome probability.
For X specified outcomes in n independent trials with a common probability p, X/n estimates p. For failures, p refers to a defined horizon. When X = 0, inverting P(X = 0) = (1 - p)^n gives the exact one-sided upper bound p_U = 1 - alpha^(1/n). At 95% confidence alpha = 0.05; the 3/n approximation is less accurate for small n. This endpoint calculation needs no constant-hazard assumption. [1] [2] [3]
A drive known to survive the planned horizon has a complete binary outcome even though its lifetime is right-censored. Early loss before that horizon leaves the outcome unknown. Repeated daily records from one drive are not additional independent trials. Different horizons or dependent failures need an observation model that accounts for those differences. A proportion of drives failing is not failures per unit exposure. [4] [5]
Relations
- related
- confidence-interval, poisson-rate-model, afr, drive-day, right-censoring