• Journal of Preventive Medicine and Public Health
• /
• v.31 no.4 s.63
• /
• pp.875-884
• /
• 1998

Missing observations are common in medical research and health survey research. Several statistical methods to handle the missing data problem have been proposed. The EM algorithm (Expectation-Maximization algorithm) is one of the ways of efficiently handling the missing data problem based on sufficient statistics. In this paper, we developed statistical models and methods for survey data with multivariate missing observations. Especially, we adopted the EM algorithm to handle the multivariate missing observations. We assume that the multivariate observations follow a multivariate normal distribution, where the mean vector and the covariance matrix are primarily of interest. We applied the proposed statistical method to analyze data from a health survey. The data set we used came from a physician survey on Resource-Based Relative Value Scale(RBRVS). In addition to the EM algorithm, we applied the complete case analysis, which uses only completely observed cases, and the available case analysis, which utilizes all available information. The residual and normal probability plots were evaluated to access the assumption of normality. We found that the residual sum of squares from the EM algorithm was smaller than those of the complete-case and the available-case analyses.

• Journal of Intelligence and Information Systems
• /
• v.9 no.2
• /
• pp.51-63
• /
• 2003

Collaborative filtering is one of the methodologies that are most widely used for recommendation system. It is based on a data matrix of each customer's preferences of products. There could be a lot of missing values in such preference data matrix. This incomplete data is one of the reasons to deteriorate the accuracy of recommendation system. There are several treatments to deal with the incomplete data problem such as case deletion and single imputation. Those approaches are simple and easy to implement but they may provide biased results. Multiple imputation method imputes m values for each missing value. It overcomes flaws of single imputation approaches through considering the uncertainty of missing values. The objective of this paper is to suggest multiple imputation-based collaborative filtering approach for recommendation system to improve the accuracy in prediction performance. The experimental works show that the proposed approach provides better performance than the traditional Collaborative filtering approach, especially in case that there are a lot of missing values in dataset used for recommendation system.

• Survey Research
• /
• v.9 no.2
• /
• pp.1-27
• /
• 2008

Problems of incomplete data are pervasive in statistical analysis. In particular, incomplete data have been an important challenge in repeated measures studies. The objective of this study is to give a brief introduction to missing data mechanisms and conventional/recent missing data methods and to assess the performance of various missing data methods under ignorable and non-ignorable missingness mechanisms. Given the inadequate attention to longitudinal studies with missing data, this study applied recent advances in missing data methods to repeated measures models and investigated the performance of various missing data methods, such as FIML (Full Information Maximum Likelihood Estimation) and MICE(Multivariate Imputation by Chained Equations), under MCAR, MAR, and MNAR mechanisms. Overall, the results showed that listwise deletion and mean imputation performed poorly compared to other recommended missing data procedures. The better performance of EM, FIML, and MICE was more noticeable under MAR compared to MCAR. With the non-ignorable missing data, this study showed that missing data methods did not perform well. In particular, this problem was noticeable in slope-related estimates. Therefore, this study suggests that if missing data are suspected to be non-ignorable, developmental research may underestimate true rates of change over the life course. This study also suggests that bias from non-ignorable missing data can be substantially reduced by considering rich information from variables related to missingness.

• Journal of the Korea Convergence Society
• /
• v.12 no.8
• /
• pp.39-45
• /
• 2021

This paper discusses how to handle incomplete data including missing values. Optimally processing the missing value means obtaining an estimate that is the closest to the original value from the information contained in the training data, and replacing the missing value with this value. The way to achieve this is to use a decision tree that is completed in the process of classifying information by the classifier. In other words, this decision tree is obtained in the process of learning by inputting only complete information that does not include loss values among all training data into the C4.5 classifier. The nodes of this decision tree have classification variable information, and the higher node closer to the root contains more information, and the leaf node forms a classification region through a path from the root. In addition, the average of classified data events is recorded in each region. Events including the missing value are input to this decision tree, and the region closest to the event is searched through a traversal process according to the information of each node. The average value recorded in this area is regarded as an estimate of the missing value, and the compensation process is completed.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.14 no.1
• /
• pp.53-62
• /
• 2010

This paper introduces a classification algorithm which can be applied to a learning problem with incomplete data sets, missing variable values or a class value. This algorithm uses a data expansion method which utilizes weighted values and probability techniques. It operates by extending a classifier which are considered to be in the optimal projection plane based on Fisher's formula. To do this, some equations are derived from the procedure to be applied to the data expansion. To evaluate the performance of the proposed algorithm, results of different measurements are iteratively compared by choosing one variable in the data set and then modifying the rate of missing and non-missing values in this selected variable. And objective evaluation of data sets can be achieved by comparing, the result of a data set with non-missing variable with that of C4.5 which is a known knowledge acquisition tool in machine learning.

• KIPS Transactions on Software and Data Engineering
• /
• v.8 no.2
• /
• pp.67-78
• /
• 2019

Accurate electric load forecasting is very important in the efficient operation of the smart grid. Recently, due to the development of IT technology, many works for constructing accurate forecasting models have been developed based on big data processing using artificial intelligence techniques. These forecasting models usually utilize external factors such as temperature, humidity and historical electric load as independent variables. However, due to diverse internal and external factors, historical electrical load contains many missing data, which makes it very difficult to construct an accurate forecasting model. To solve this problem, in this paper, we propose a random forest-based missing data recovery scheme and construct an electric load forecasting model based on multilayer perceptron using the estimated values of missing data and external factors. We demonstrate the performance of our proposed scheme via various experiments.

• Journal of the Korea Convergence Society
• /
• v.12 no.11
• /
• pp.91-97
• /
• 2021

This paper uses a data extension technique originally designed for the rule refinement problem to handling incomplete data. This technique is characterized in that each event can have a weight indicating importance, and each variable can be expressed as a probability value. Since the key problem in this paper is to find the probability that is closest to the missing value and replace the missing value with the probability, three different algorithms are used to find the probability for the missing value and then store it in this data structure format. And, after learning to classify each information area with the SVM classification algorithm for evaluation of each probability structure, it compares with the original information and measures how much they match each other. The three algorithms for the imputation probability of the missing value use the same data structure, but have different characteristics in the approach method, so it is expected that it can be used for various purposes depending on the application field.

• Journal of the Korean Data and Information Science Society
• /
• v.23 no.6
• /
• pp.1183-1194
• /
• 2012

When it comes to a survey, no answer would occur frequently. Therefore various methods for handling nonresponse have been applied to analyse the survey. In this paper, the ratio of occurrence of two type of nonresponse cases - unit nonresponse and item nonresponse - is presented using previous real survey data, and we compared complete data and data with nonresponse. We suggest the reason of happening of nonresponse and the ratio of nonresponse using data collected through group interviews.

• Communications for Statistical Applications and Methods
• /
• v.13 no.1
• /
• pp.101-111
• /
• 2006

One of the most widely used method of handling missingness in microarray data is the kNN(k Nearest Neighborhood) method. Recently Li and Gui (2004) suggested, so called PCR(Partial Cox Regression) method which deals with censored survival times and microarray data efficiently via kNN imputation method. In this article, we try to show that the way to treat missingness eventually affects the further statistical analysis.

• The Korean Journal of Applied Statistics
• /
• v.28 no.2
• /
• pp.159-174
• /
• 2015

Dragomiretskiy and Zosso (2014) developed a new decomposition method, termed variational mode decomposition (VMD), which is efficient for handling the tone detection and separation of signals. However, VMD may be inefficient in the presence of missing data since it is based on a fast Fourier transform (FFT) algorithm. To overcome this problem, we propose a new approach based on a novel combination of VMD and hierarchical (or h)-likelihood method. The h-likelihood provides an effective imputation methodology for missing data when VMD decomposes the signal into several meaningful modes. A simulation study and real data analysis demonstrates that the proposed method can produce substantially effective results.