• International Journal of Computer Science & Network Security
• /
• v.21 no.12spc
• /
• pp.549-555
• /
• 2021

Machine Learning(ML) splits data into 3 parts, which are usually 60% for training, 20% for validation, and 20% for testing. It just splits quantitatively instead of selecting each set of data by a criterion, which is very important concept for the adequacy of test data. ML measures a model's accuracy by applying a set of validation data, and revises the model until the validation accuracy reaches on a certain level. After the validation process, the complete model is tested with the set of test data, which are not seen by the model yet. If the set of test data covers the model's attributes well, the test accuracy will be close to the validation accuracy of the model. To make sure that ML's set of test data works adequately, we design an experiment and see if the test accuracy of model is always close to its validation adequacy as expected. The experiment builds 100 different SVM models for each of six data sets published in UCI ML repository. From the test accuracy and its validation accuracy of 600 cases, we find some unexpected cases, where the test accuracy is very different from its validation accuracy. Consequently, it is not always true that ML's set of test data is adequate to assure a model's quality.

• Communications for Statistical Applications and Methods
• /
• v.12 no.2
• /
• pp.359-364
• /
• 2005

This paper considers a specification test of conditional Poisson regression model for time series count data. Although conditional models for count data have received attention and proposed in several ways, few studies focused on checking its adequacy. Motivated by the test of martingale difference assumption, a specification test via Ljung-Box statistic is proposed in the conditional model of the time series count data. In order to illustrate the performance of Ljung- Box test, simulation results will be provided.

• Journal of the Korean Data and Information Science Society
• /
• v.27 no.2
• /
• pp.559-564
• /
• 2016

The Durbin-Watson (DW) test in regression model and the Ljung-Box (LB) test in ARMA (autoregressive moving average) model are typical examples of correlated error tests. The DW test is used for detecting autocorrelation of errors using the residuals from a regression analysis. The LB test is used for specifying the correct ARMA model using the first some sample autocorrelations based on the residuals of a tted ARMA model. In this article, simulations with four data generating processes have been carried out to evaluate their performances as correlated error tests. Our simulations show that the DW test is severely dependent on the assumed AR(1) model but isn't sensitive enough to reject the misspecified model and that the LB test reports lackluster performance in general.

• Communications for Statistical Applications and Methods
• /
• v.16 no.2
• /
• pp.325-334
• /
• 2009

In this research, we propose a nonparametric test procedure for the right censored and grouped data under the additive hazards model. For deriving the test statistics, we use the likelihood principle. Then we illustrate proposed test with an example and compare the performance with other procedure by obtaining empirical powers. Finally we discuss some interesting features concerning the proposed test.

• Journal of the Korean Data and Information Science Society
• /
• v.24 no.6
• /
• pp.1551-1559
• /
• 2013

In this paper, we analyze the Korean Won/Japanese 100 Yen exchange rate data based on the ARMA-GARCH model, and perform the test for detecting the parameter changes. As a test statistics, we employ the cumulative sum (CUSUM) test for ARMA-GARCH model, which is introduced by Lee and Song (2008). Our empirical analysis indicates that the KRW/JPY exchange rate series experienced several parameter changes during the period from January 2000 to December 2012, which leads to a fitting of AR-IGARCH model to the whole series.

• 한국기계연구소 소보
• /
• s.9
• /
• pp.183-191
• /
• 1982

This report describes the analysis method of the full-scale propulsive performance by using the data of model test and the full-scale speed trial. The model test data were analyzed by the computer program "PPTT" based on "1978 ITTC Performance Prediction Method for Single Screw Ships." Also the full-scale speed trial data were analyzed by the computer program "SSTT" based on the newly proposed “SRS-KIMM Standard Method of Speed Trial Analysis." An analysis of model and full-scale test data was carried out for a 60.000 DWT Bulk Carrier and the correlation between model and full-scale ship was stuied.

• Journal of the Korean Data and Information Science Society
• /
• v.18 no.1
• /
• pp.11-16
• /
• 2007

Since the seminal work of Engle (1982), many researchers and practitioners have developed ARCH-type models to deal with volatility modelling, which, for instance, is crucial to perform the task of derivative pricing, measuring risk, and risk hedging. In this paper, we base the GARCH(1,1) model to analyze the KOSPI200 data, and perform the CUSUM test for detecting parameter changes in the GARCH model. It is shown that the data suffers from a parameter change.

• Journal of the Korean Data and Information Science Society
• /
• v.15 no.1
• /
• pp.31-39
• /
• 2004

In this paper, we consider two components system which the lifetimes follow bivariate pareto model with bivariate random censored data. We assume that the censoring times are independent of the lifetimes of the two components. We develop large sample test for testing independence between two components. Also we present a simulation study which is the test based on asymptotic normal distribution in testing independence.

• Journal of the Korean Data and Information Science Society
• /
• v.15 no.1
• /
• pp.129-139
• /
• 2004

We consider the problem of testing for outliers in generalized linear model. We proceed by first specifying a mean shift outlier model, assuming the suspect set of ourliers is known. Given this model, we discuss standard approaches to obtaining score test for outliers as an alternative to the likelihood ratio test.

• Journal of the Korean Data and Information Science Society
• /
• v.14 no.4
• /
• pp.951-963
• /
• 2003

In this paper, we propose goodness of fit test statistics based on the estimated Kullback-Leibler information functions using the data from three step stress accelerated life test. This acceleration model is assumed to be a tampered random variable model. The power of the proposed test under various alternatives is compared with Kolmogorov-Smirnov statistic, Cramer-von Mises statistic and Anderson-Darling statistic.