• Journal of the Institute of Electronics and Information Engineers
• /
• v.53 no.3
• /
• pp.99-106
• /
• 2016

In this paper, we will discuss about correction method of missing data on noisy observation matrix and uncertainty analysis for the potential noise. In situations without missing data in an observation matrix, this solution is known to be accurately induced by SVD (Singular Value Decomposition). However, usually the several entries of observation matrix have not been observed and other entries have been perturbed by the influence of noise. In this case, it is difficult to find the solution as well as cause the 3D reconstruction error. Therefore, in order to minimize the 3D reconstruction error, above all things, it is necessary to correct reliably the missing data under noise distribution and to give a quantitative evaluation for the corrected results. This paper focuses on a method for correcting missing data using geometrical properties between 2D projected object and 3D reconstructed shape and for estimating a noise level of the observation matrix using ranks of SVD in order to quantitatively evaluate the performance of the correction algorithm.

• Journal of the Korean Statistical Society
• /
• v.18 no.1
• /
• pp.13-25
• /
• 1989

This paper suggests a classification rule with unequal covariance matrices when a patterned incomplete data are involved in the discriminant analysis. This is an extension of Geisser's (1966) result to the case of missing observations. For the calssificaiton rule, we introduce an algorithm which contains data augmentation step and Monte Carlo integration step and show that the algorithm yields a consistant estimator of true classification probability. The proposed method is compared to the complete observation vector method through a Monte Carlo study. The results show that the suggested method, in general, performs better than the complete observation vector method which ignores those vectors of observation with one or more missing values from the analysis. The results also verify the consistency of the algorithm.

• Communications for Statistical Applications and Methods
• /
• v.16 no.4
• /
• pp.675-686
• /
• 2009

In this paper, it will be assumed that there are two distinct populations which are multivariate normal with equal covariance matrix. We also assume that the two populations are equally likely and the costs of misclassification are equal. The classification rule depends on the situation when the training samples include missing values or not. We consider the bootstrap confidence intervals for classification error rate when a block of observation is missing.

• Ocean and Polar Research
• /
• v.44 no.2
• /
• pp.179-190
• /
• 2022

Data missing in optical satellite images caused by natural variations have been a crucial barrier in observing the status of marine surfaces. Although there have been many attempts to fill the gaps of non-observation, there is little research to analyze the ratio of missing grids to overall sea grids and their seasonal patterns. This report introduces the method of quantifying the distribution of missing points and then shows how the missing points have spatial correlation and seasonal trends. Both temporal and spatial integration methods are compared to assess the effectiveness of reducing missing data. The temporal integration shows more outstanding performance than the spatial integration. Moran's I and K-function with statistical hypothesis testing show that missing grids are clustered and there is a non-random distribution from daily integration. The result of the seasonality test for Moran's I through a periodogram shows dependency on full-year, half-year, and quarter-year periods respectively. These analysis results can be used to deduce appropriate integration periods with permissible estimation errors.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.34 no.4
• /
• pp.109-118
• /
• 2022

The data of the missing section among the vertex surface sea temperature observation data was imputed using the Bidirectional Recurrent Neural Network(BiRNN). Among artificial intelligence techniques, Recurrent Neural Networks (RNNs), which are commonly used for time series data, only estimate in the direction of time flow or in the reverse direction to the missing estimation position, so the estimation performance is poor in the long-term missing section. On the other hand, in this study, estimation performance can be improved even for long-term missing data by estimating in both directions before and after the missing section. Also, by using all available data around the observation point (sea surface temperature, temperature, wind field, atmospheric pressure, humidity), the imputation performance was further improved by estimating the imputation data from these correlations together. For performance verification, a statistical model, Multivariate Imputation by Chained Equations (MICE), a machine learning-based Random Forest model, and an RNN model using Long Short-Term Memory (LSTM) were compared. For imputation of long-term missing for 7 days, the average accuracy of the BiRNN/statistical models is 70.8%/61.2%, respectively, and the average error is 0.28 degrees/0.44 degrees, respectively, so the BiRNN model performs better than other models. By applying a temporal decay factor representing the missing pattern, it is judged that the BiRNN technique has better imputation performance than the existing method as the missing section becomes longer.

• Journal of the Korean Data and Information Science Society
• /
• v.23 no.6
• /
• pp.1271-1277
• /
• 2012

In this paper, we propose an estimation method of the parameter in an exponential distribution based on a progressive Type I interval censored sample with semi-missing observation. The maximum likelihood estimator (MLE) of the parameter in the exponential distribution cannot be obtained explicitly because the intervals are not equal in length under the progressive Type I interval censored sample with semi-missing data. To obtain the MLE of the parameter for the sampling scheme, we propose a method by which progressive Type I interval censored sample with semi-missing data is converted to the progressive Type II interval censored sample. Consequently, the estimation procedures in the progressive Type II interval censored sample can be applied and we obtain the MLE of the parameter and survival function. It will be shown that the obtained estimators have good performance in terms of the mean square error (MSE) and mean integrated square error (MISE).

• Journal of Integrative Natural Science
• /
• v.5 no.2
• /
• pp.135-138
• /
• 2012

The reduction of efficiency of missing observations on complete diallel cross designs are examined. we studies robustness of optimal block designs for estimating general combining ability against loss of missing observations in diallel cross. A-efficiencies suggest that these designs are fairly robust. Simple g-inverses may be found for the information matrices of the line effects which allow evaluation of expressions for the variances of the differences between the pairs of line effects with missing observations. we numerically calculate the reduction of efficiency for estimating general combining ability against loss of missing observations in diallel cross.

• The korean journal of orthodontics
• /
• v.50 no.6
• /
• pp.407-417
• /
• 2020

Objective: To investigate the dental phenotypes and treatment modalities (Tx-Mod) in Korean patients with Parry-Romberg syndrome (PRS) using longitudinal data. Methods: The samples consisted of 10 PRS patients, who were treated and/or followed-up at Seoul National University Dental Hospital between 1998 and 2019. Using a novel PRS severity index based on the numbers of the atrophy-involved area and asymmetry-involved item, we classified them into mild (n = 3), moderate (n = 2), and severe (n = 5). Dental phenotypes, including congenitally missing tooth (Con-Missing-Tooth), microdontia, tooth with short root (Short-Root), tooth with dilacerated root, and delayed eruption/impacted tooth, were investigated along with Tx-Mod. Results: The side of occurrence of all dental phenotypes showed 100% concordance with the side of PRS involvement. The most two common dental phenotypes were Con-Missing-Tooth and Short-Root (n = 29 and n = 17 in six patients). The sums of the average number of Con-Missing-Tooth and Short-Root increased from mild PRS to moderate PRS and severe PRS cases (1.0, 6.0, and 6.2). In terms of Tx-Mod, growth observation due to mild atrophy, fixed orthodontic treatment, and grafting were used for mild PRS cases. Tx-Mod for moderate PRS cases involved growth observation for surgery due to an early age at the initial visit. For severe PRS cases, diverse Tx-Mod combinations including unilateral functional appliance, fixed orthodontic treatment, growth observation, grafting, and orthognathic surgery were used. Conclusions: The novel PRS severity index may be useful to provide primary data for individualized diagnosis and treatment planning for PRS patients.

• Communications for Statistical Applications and Methods
• /
• v.4 no.1
• /
• pp.49-63
• /
• 1997

We consider the problem of classifying a p x 1 observation into one of two multivariate normal populations when the training smaples contain a block of missing observation. A new classification procedure is proposed which is a linear combination of two discriminant functions, one based on the complete samples and the other on the incomplete samples. The new discriminant function is easy to use.

• 한국데이터정보과학회:학술대회논문집
• /
• 2006.11a
• /
• pp.144-144
• /
• 2006