• Communications for Statistical Applications and Methods
• /
• v.16 no.5
• /
• pp.803-812
• /
• 2009

The application of extreme value theory to financial data is a fairly recent innovation. The classical annual maximum method is to fit the generalized extreme value distribution to the annual maxima of a data series. An alterative modern method, the so-called threshold method, is to fit the generalized Pareto distribution to the excesses over a high threshold from the data series. A more substantial variant is to take the point-process viewpoint of high-level exceedances. That is, the exceedance times and excess values of a high threshold are viewed as a two-dimensional point process whose limiting form is a non-homogeneous Poisson process. In this paper, we apply the two-dimensional non-homogeneous Poisson process model to daily losses, daily negative log-returns, in the data series of KBW/USD exchange rate, collected from January 4th, 1982 until December 31 st, 2008. The main question is how to estimate extreme quantiles of losses such as the 10-year or 50-year return level.

• The Korean Journal of Applied Statistics
• /
• v.25 no.1
• /
• pp.101-114
• /
• 2012

In car insurance, the loss ratio is the ratio of total losses paid out in claims divided by the total earned premiums. In order to minimize the loss to the insurance company, estimating extreme quantiles of loss ratio distribution is necessary because the loss ratio has essential prot and loss information. Like other types of insurance related datasets, the distribution of the loss ratio has heavy-tailed distribution. The Peaks over Threshold(POT) and the Hill estimator are commonly used to estimate extreme quantiles for heavy-tailed distribution. This article compares and analyzes the performances of various kinds of parameter estimating methods by using a simulation and the real loss ratio of car insurance data. In addition, we estimate extreme quantiles using the Hill estimator. As a result, the simulation and the loss ratio data applications demonstrate that the POT method estimates quantiles more accurately than the Hill estimation method in most cases. Moreover, MLE, Zhang, NLS-2 methods show the best performances among the methods of the GPD parameters estimation.

• The Korean Journal of Applied Statistics
• /
• v.27 no.4
• /
• pp.655-665
• /
• 2014

An adequate understanding and response to natural hazards such as heat wave, heavy rainfall and severe drought is required. We apply extreme value theory to analyze these abnormal weather phenomena. It is common for extremes in climatic data to be nonstationary in space and time. In this paper, we analyze summer rainfall data in South Korea using exceedance values over thresholds estimated by quantile regression with location information and time as covariates. We group weather stations in South Korea into 5 clusters and t extreme value models to threshold exceedances for each cluster under the assumption of independence in space and time as well as estimates of uncertainty for spatial dependence as proposed in Northrop and Jonathan (2011).

• The Korean Journal of Applied Statistics
• /
• v.24 no.5
• /
• pp.833-845
• /
• 2011

This paper conducts a statistical analysis of extreme values for both daily log-returns and daily negative log-returns, which are computed using a collection of KOSPI data from January 3, 1998 to August 31, 2011. The Poisson-GPD model is used as a statistical analysis model for extreme values and the maximum likelihood method is applied for the estimation of parameters and extreme quantiles. To the Poisson-GPD model is also added the Bayesian method that assumes the usual noninformative prior distribution for the parameters, where the Markov chain Monte Carlo method is applied for the estimation of parameters and extreme quantiles. According to this analysis, both the maximum likelihood method and the Bayesian method form the same conclusion that the distribution of the log-returns has a shorter right tail than the normal distribution, but that the distribution of the negative log-returns has a heavier right tail than the normal distribution. An advantage of using the Bayesian method in extreme value analysis is that there is nothing to worry about the classical asymptotic properties of the maximum likelihood estimators even when the regularity conditions are not satisfied, and that in prediction it is effective to reflect the uncertainties from both the parameters and a future observation.

• The Korean Journal of Applied Statistics
• /
• v.33 no.6
• /
• pp.753-762
• /
• 2020

With an increasing interest in integrating artificial intelligence (AI) into interview processes, the Republic of Korea (ROK) army is trying to lead and analyze AI-powered interview platform. This study is to analyze the AI interview data using a unified non-crossing multiple quantile tree (UNQRT) model. Compared to the UNQRT, the existing models, such as quantile regression and quantile regression tree model (QRT), are inadequate for the analysis of AI interview data. Specially, the linearity assumption of the quantile regression is overly strong for the aforementioned application. While the QRT model seems to be applicable by relaxing the linearity assumption, it suffers from crossing problems among estimated quantile functions and leads to an uninterpretable model. The UNQRT circumvents the crossing problem of quantile functions by simultaneously estimating multiple quantile functions with a non-crossing constraint and is robust from extreme quantiles. Furthermore, the single tree construction from the UNQRT leads to an interpretable model compared to the QRT model. In this study, by using the UNQRT, we explored the relationship between the results of the Army AI interview system and the existing personnel data to derive meaningful results.