A0321
Title: Prediction of order statistics based on ordered generalized ranked set sampling
Authors: Masato Kitani - Tokyo University of Science (Japan) [presenting]
Katsuyuki Yuasa - Tokyo University of Science (Japan)
Hidetoshi Murakami - Tokyo University of Science (Japan)
Abstract: In statistical inference, a prediction interval is important for estimating an interval in which a future observation will fall. The distribution-free prediction intervals for order statistics of future observations have been introduced previously. In practical analysis, the prediction of future extreme observations plays an important role, such as the largest river flow in the next few years. However, large sample sizes are needed to construct the prediction intervals with sufficient coverage probability for future extreme order statistics. Therefore, we propose a prediction interval using the generalized ranked set sampling, which can observe only selected order statistics by using the rank of the data before observation. We show that the proposed prediction interval has sufficient coverage probabilities with small sample sizes. Furthermore, we show that the proposed prediction interval has desirable properties compared to other existing methods by numerical simulations.
