• Communications for Statistical Applications and Methods
• /
• 제30권2호
• /
• pp.215-226
• /
• 2023

Deep learning has made great strides in the field of unstructured data such as text, images, and audio. However, in the case of tabular data analysis, machine learning algorithms such as ensemble methods are still better than deep learning. To keep up with the performance of machine learning algorithms with good predictive power, several deep learning methods for tabular data have been proposed recently. In this paper, we review the latest deep learning models for tabular data and compare the performances of these models using several datasets. In addition, we also compare the latest boosting methods to these deep learning methods and suggest the guidelines to the users, who analyze tabular datasets. In regression, machine learning methods are better than deep learning methods. But for the classification problems, deep learning methods perform better than the machine learning methods in some cases.

• 스마트미디어저널
• /
• 제13권4호
• /
• pp.9-15
• /
• 2024

Tree-based algorithms have been the dominant methods used build a prediction model for tabular data. This also includes personal credit data. However, they are limited to compatibility with categorical and numerical data only, and also do not capture information of the relationship between other features. In this work, we proposed an ensemble model using the Transformer architecture that includes text features and harness the self-attention mechanism to tackle the feature relationships limitation. We describe a text formatter module, that converts the original tabular data into sentence data that is fed into FinBERT along with other text features. Furthermore, we employed FT-Transformer that train with the original tabular data. We evaluate this multi-modal approach with two popular tree-based algorithms known as, Random Forest and Extreme Gradient Boosting, XGBoost and TabTransformer. Our proposed method shows superior Default Recall, F1 score and AUC results across two public data sets. Our results are significant for financial institutions to reduce the risk of financial loss regarding defaulters.

• 사물인터넷융복합논문지
• /
• 제9권6호
• /
• pp.119-124
• /
• 2023

최근 딥러닝은 다양한 분야에서 전통적인 기계학습에 비해 월등히 높은 성능을 보이고 있으며, 패턴인식을 위한 보편적인 방법으로 자리 잡아 가고 있다. 하지만, 이에 비해 정형데이터를 사용하는 분류 문제에서는 여전히 머신러닝 기법이 주류를 이루고 있다. 본 논문에서는 정형데이터를 고차원 텐서로 변환하는 네트워크 모듈을 제안하며, 이 모듈을 보편적인 딥러닝 네트워크와 함께 구성하여 정형데이터의 분류 문제에 적용하였다. 제안된 방법은 4종의 데이터셋을 활용하여 학습 및 검증되었으며, 제안된 방법은 90.22%의 평균 정확도를 달성하여, 최신 딥러닝 모델인 TabNet에 비해 2.55%p 높은 정확도를 보였다. 제안된 방법은 컴퓨터 비전 분야에서 높은 성능을 보이는 다양한 네트워크 구조를 정형데이터에 활용할 수 있다는 점에서 의미가 있다.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2022년도 추계학술대회
• /
• pp.164-167
• /
• 2022

표 기계독해에서는 도메인에 따라 언어모형에 필요한 지식이나 표의 구조적인 형태가 변화하면서 텍스트 데이터에 비해서 더 큰 성능 하락을 보인다. 본 논문에서는 표 기계독해에서 이러한 도메인의 변화에 강건한 사전학습 표 언어 모형 구축을 위한 의미있는 표 데이터 선별을 통한 사전학습 데이터 구축 방법과 적대적인 학습 방법을 제안한다. 추출한 표 데이터에서 구조적인 정보가 없이 웹 문서의 장식을 위해 사용되는 표 데이터 검출을 위해 Heuristic을 통한 규칙을 정의하여 HEAD 데이터를 식별하고 표 데이터를 선별하는 방법을 적용했으며, 구조적인 정보를 가지는 일반적인 표 데이터와 엔티티에 대한 지식 정보를 가지는 인포박스 데이터간의 적대적 학습 방법을 적용했다. 기존의 정제되지 않는 데이터로 학습했을 때와 비교하여 데이터를 정제하였을 때, KorQuAD 표 데이터에서 F1 3.45, EM 4.14가 증가하였으며, Spec 표 질의응답 데이터에서 정제하지 않았을 때와 비교하여 F1 19.38, EM 4.22가 증가한 성능을 보였다.

• ETRI Journal
• /
• 제42권6호
• /
• pp.899-911
• /
• 2020

The data in tables are accurate and rich in information, which facilitates the performance of information extraction and question answering (QA) tasks. TableQA, which is based on tables, solves problems by understanding the table structure and searching for answers to questions. In this paper, we introduce both novice and intermediate Korean TableQA tasks that involve deducing the answer to a question from structured tabular data and using it to build a question answering pair. To solve Korean TableQA tasks, we use S3-NET, which has shown a good performance in machine reading comprehension (MRC), and propose a method of converting structured tabular data into a record format suitable for MRC. Our experimental results show that the proposed method outperforms a baseline in both the novice task (exact match (EM) 96.48% and F1 97.06%) and intermediate task (EM 99.30% and F1 99.55%).

• 응용통계연구
• /
• 제23권2호
• /
• pp.325-333
• /
• 2010

Protection against disclosure of survey respondents' identifiable and/or sensitive information is a prerequisite for statistical agencies that release microdata files from their sample surveys. Coarsening is one of popular methods for protecting the confidentiality of the data. Grouped data can be released in the form of microdata or tabular data. Instead of releasing the data in a tabular form only, having microdata available to the public with interval codes with their representative values greatly enhances the utility of the data. It allows the researchers to compute covariance between the variables and build statistical models or to run a variety of statistical tests on the data. It may be conjectured that the variance of the interval data is lower that of the ungrouped data in the sense that the coarsened data do not have the within interval variance. This conjecture will be investigated using the uniform and triangular distributions. Traditionally, midpoint is used to represent all the values in an interval. This approach implicitly assumes that the data is uniformly distributed within each interval. However, this assumption may not hold, especially in the last interval of the economic data. In this paper, we will use three distributional assumptions - uniform, Pareto and lognormal distribution - in the last interval and use either midpoint or median for other intervals for wage and food costs of the Statistics Korea's 2006 Household Income and Expenditure Survey(HIES) data and compare these approaches in terms of the first two moments.

• 한국농공학회논문집
• /
• 제56권6호
• /
• pp.93-102
• /
• 2014

Generally data loading is a laborous job to develop models. Meteorological data is basic input data for hydrological models, it is provided through websites of Korea Meteorological Administration (KMA). The website of KMA provides daily meteorological observation data with tabular format classified by years, items, stations. It is cumbersome to manipulate tabular format for model inputs such as time series and multi-item or multi-station data. The provider oriented services which broadcast restricted formed information have caused inconvenient processes. Tim O'Reilly introduces "Web 2.0" which focuses on providing a service based on data. The top ranked IT companies such as google, yahoo, daum, and naver provide customer oriented services with Open API (Application Programming Interface). A RESTful web service, typical implementation for Open API, consists URI request and HTTP response which are simple and light weight protocol than SOAP (Simple Object Access Protocol). The aim of this study is to develop a web-based service that helps loading data for human use instead of machine use. In this study, the developed RESTful web service provides Open API for manipulating meteorological data. The proposed Open API can easily access from spreadsheet programs, web browsers, and various programming environments.

• Geomechanics and Engineering
• /
• 제30권3호
• /
• pp.247-258
• /
• 2022

A field velocity resistivity probe (FVRP) can measure compressional waves, shear waves and electrical resistivity in boreholes. The objective of this study is to perform the soil classification through a machine learning technique through elastic wave velocity and electrical resistivity measured by FVRP. Field and laboratory tests are performed, and the measured values are used as input variables to classify silt sand, sand, silty clay, and clay-sand mixture layers. The accuracy of k-nearest neighbors (KNN), naive Bayes (NB), random forest (RF), and support vector machine (SVM), selected to perform classification and optimize the hyperparameters, is evaluated. The accuracies are calculated as 0.76, 0.91, 0.94, and 0.88 for KNN, NB, RF, and SVM algorithms, respectively. To increase the amount of data at each soil layer, the synthetic minority oversampling technique (SMOTE) and conditional tabular generative adversarial network (CTGAN) are applied to overcome imbalance in the dataset. The CTGAN provides improved accuracy in the KNN, NB, RF and SVM algorithms. The results demonstrate that the measured values by FVRP can classify soil layers through three kinds of data with machine learning algorithms.

• 한국정보통신학회논문지
• /
• 제26권1호
• /
• pp.64-69
• /
• 2022

공공기술 사업화는 정부가 주도하는 과학기술의 혁신과 R&D 성과를 민간에 이전하는 것으로 경제 성장을 주도하는 핵심 성과로 인식되고 있다. 따라서 기술 이전을 활성화시키기 위해 성공 요인을 식별하거나 사업화 가능성이 높은 공공기술과 수요기업을 매칭하는 다양한 기계학습의 방법들이 연구되고 있다. 하지만 공공기술 사업화 데이터는 표 형태로 구성되어 있고, 성공-실패 비율이 큰 차이를 보이는 불균형 상태이기 때문에 기계학습 성능이 높지 않는 문제점을 가지고 있다. 이 논문에서는 표 형태로 구성된 공공기술 데이터에서 불균형을 해소하기 위해 CTGAN을 활용하는 방법을 제시한다. 또한 제시된 방법의 효과를 검증하기 위해 실제 공공기술 사업화 데이터를 활용하여 통계적 접근방법인 SMOTE와 비교 실험을 수행하였다. 다수의 실험 사례에서 CTGAN은 공공기술 사업화 성공사례를 안정적으로 예측하는 것을 확인하였다.

• Journal of information and communication convergence engineering
• /
• 제19권4호
• /
• pp.228-233
• /
• 2021

In this study, we analyze the credit information (loan, delinquency information, etc.) of individual business owners to generate voluminous training data to establish a bankruptcy prediction model through a partial synthetic training technique. Furthermore, we evaluate the prediction performance of the newly generated data compared to the actual data. When using conditional tabular generative adversarial networks (CTGAN)-based training data generated by the experimental results (a logistic regression task), the recall is improved by 1.75 times compared to that obtained using the actual data. The probability that both the actual and generated data are sampled over an identical distribution is verified to be much higher than 80%. Providing artificial intelligence training data through data synthesis in the fields of credit rating and default risk prediction of individual businesses, which have not been relatively active in research, promotes further in-depth research efforts focused on utilizing such methods.