Confidence interval estimation using standardized time series. (English) Zbl 0532.62067
In this paper the author studies a novel way to form estimates of the confidence interval for above all the mean of a time series. The basic idea is to normalize or standardize the time series by, roughly speaking, centering and scaling the series, and also changing the time scale to the unit interval. For such series the limiting distribution is a Wiener process, which result gives a handle to forming the confidence interval estimates.
The paper is clearly written and all that not technical. It includes results from computer simulations, which compare favorably with the results of traditional estimation techniques. Nonetheless, to the casual reader the very point in the proposed idea, namely, the advantage of the standardization, remains a bit puzzling.
The paper is clearly written and all that not technical. It includes results from computer simulations, which compare favorably with the results of traditional estimation techniques. Nonetheless, to the casual reader the very point in the proposed idea, namely, the advantage of the standardization, remains a bit puzzling.
Reviewer: J.Rissanen
MSC:
62M10 | Time series, auto-correlation, regression, etc. in statistics (GARCH) |
62F25 | Parametric tolerance and confidence regions |