A Study on Credit Default Prediction using eXplainable Artificial Intelligence
Abstract : This study aimed to empirically compare the performance of the logistic regression model and the XGBoost model for predicting individual credit card default, and to enhance the interpretability of the models through eXplainable Artificial Intelligence (XAI). Using American Express dataKaggle, I analyzed feature importance through coefficient-based, gain-based, and SHAP (SHapley Additive exPlanations)–based methods with the following results. First, while both models demonstrated excellent predictive performance, XGBoost achieved zero False Negatives (Type II error) in the confusion matrix analysis, thereby perfectly identifying all actual default customers. This has important practical implications in risk management as it helps prevent financial losses for financial institutions. Second, the SHAP analysis confirmed that credit score, previous defaults, and credit limit usage are key variables in predicting default. SHAP, based on Shapley values, clearly decomposes the contribution of each variable to individual predictions while considering interactions among variables, leading to significantly improved interpretability compared with traditional methodologies such as coefficient-based and gain-based analysis. This provides important grounds for securing the transparency and credibility of models required in financial practice.
Keyword : Business Analytics, eXplainable Artificial Intelligence, Quantitative Finance, Financial AI, Financial Risk Management
http://dx.doi.org/10.29056/jncist.2025.12.01
Study on Radar Band Characteristics of Sinkholes through Sentinel-1 Image Preprocessing
Abstract : This study presents a ground deformation detection and early monitoring method using Sentinel-1 SAR satellite imagery to address the frequent occurrences of road subsidence and sinkholes in urban areas. Due to climate change and aging infrastructure, road collapses in cities have been increasing; however, traditional visual inspection methods are limited by time and spatial constraints, making real-time response difficult. Synthetic Aperture Radar (SAR) has the advantage of detecting ground surface changes even under conditions such as rain, cloud cover,nighttime, making it suitable for infrastructure safety monitoring. Previous studies have demonstrated the feasibility of combining SAR imagery with deep learning techniques to detect road damage, structural collapse, and ground subsidence, highlighting the effectiveness of quantitative backscatter analysis and coherence-based detection methods. In this study, Sentinel-1 SAR data were preprocessed using the SNAP platform through Deburst, Calibration, Speckle Noise Removal, and Terrain Correction steps to generate normalized intensity images. Vector layers of sinkhole and normal road areas were then constructed in QGIS, aligned with SAR data, and statistically compared in terms of backscatter intensity distribution. The analysis showed that the sinkhole area exhibited littleno backscatter signal, often appearing as discontinuous spike patterns, whereas the normal road maintained consistent intensity values. Future research should apply time-series Sentinel-1 analysis, integrate PMS (Pavement Management System) data, and develop deep learning–based change detection models to more accurately capture the spatiotemporal variations of road deformation.
Keyword : synthetic aperture radar, sinkhole, preprocessing, QGIS, detection
http://dx.doi.org/10.29056/jncist.2025.12.02
Analysis of Domestic Media Reports Related to Andong Tourism Using Bigkinds
Abstract : The purpose of this study is to analyze domestic media reports related to Andong tourism using BigKinds. In order to achieve the purpose of the study, 16,092 press reports reportedJuly 23, 2020 to July 23, 2025 were extracted and analyzed with Andong Tourism as a key search word in Bigkinds' news analysis system based on big data technology. As a result, first, the result of trend analysis of keywords related to Andong tourism was the most common at 3,574 cases in 2024. Second, as a result of analyzing Andong tourism-related words, Gyeongbuk-do, Hahoe Village, Corona 19, Gyeongju, and public transportation showed high correlation in order. Finally, as a result of analyzing the relationship related to Andong tourism, Andong, Andong-si, the Director of Culture, Tourism and Sports Bureau, UNESCO, and Dosanseowon appeared in order. This study differsprevious studies in that research was conducted through big data analysis. It was a new attempt to compare and analyze media articles related to Andong tourism, and it is believed that it will be able to help research in the social science field and other academic fields. In future research, it is necessary to design so that the title and content of the article can be analyzed together.
Keyword : Bigkinds, Andong Tourism, Media Reports, Keyword Trend, Analysis of Related Words, Analysis of The Relationship
http://dx.doi.org/10.29056/jncist.2025.12.03
Interactive System for Direct Volume Rendering in Mixed Reality Environments: An Effectiveness Study
Abstract : Direct Volume Rendering (DVR) is a technique that extracts internal structures of 3D volumetric data and effectively visualizes regions of interest by adjusting Transfer Functions (TF). Recently, as computing environments expand, research on intuitively interacting with volumetric data in Mixed Reality (MR) environments using Head-Mounted Displays (HMD) has become active. However, existing studies have relied on predefined TFsingle input methods such as hand gestures, which limited users in freely and precisely adjusting TF in immersive environments. Therefore, this study aimed to maximize control flexibility in MR environments by proposing a multi-modal interface that integrates hand gestures, eye tracking, and Large Language Model (LLM)-based voice recognition. We implemented the proposed system in the HoloLens 2 environment and analyzed the effectiveness of DVR using medical data, such as head and spine, and simulation data. Experimental results showed that while an average rendering performance of 10 FPS and expressive limitations of 1D TF were observed due to hardware constraints, intuitive data exploration and visualization adjustment were possible without complex menu manipulations through multi-modal interfaces. The results of this study present the direction and practical development potential of MR-based visualization interfaces in specialized medical and industrial fields such as surgical navigation.
Keyword : Direct Volume Rendering, Mixed Reality, Transfer Function, Visualization
http://dx.doi.org/10.29056/jncist.2025.12.04
Abstract : This study analyzes how sensory-based user experience elements influence the flow experience of users, focusing on performance pants within functional sportswear. As functional sportswear has recently expandedperformance-centered design to user experience-based design that considers sensory and emotional experiences, it has become increasingly important to clearly identify the structural relationship between sensory elements and flow experiences. In this study, tactile sensation, weight, breathability, visual simplicity, and flexibility were defined as sensory-based UX elements, and their associations with flow elements such as focus, emotional stability, time distortion, loss of self-consciousness, and sense of control were examined. A qualitative case analysis was employed, and representative productsNike, Lululemon, and Arc'teryx Veilance were selected. The analysis showed that although all three products commonly utilized sensory-based UX elements as core design strategies, each brand differed in the emphasis of sensory elements and their contributions to flow experiences. Nike enhanced focus and sense of control in active environments through lightness and stretch, while Lululemon promoted emotional stability and time distortion by minimizing garment awareness. Arc'teryx Veilance strengthened static and internalized immersion through minimal design and refined tactile qualities. These findings demonstrate that sensory-based UX elements play a critical role in forming and maintaining flow experiences in functional sportswear and suggest the potential of UX-oriented design strategies for future performance wear.
Keyword : Sensory User Experience, Flow Experience, Functional Sportswear, Performance Pants
http://dx.doi.org/10.29056/jncist.2025.12.05
A Study on the Visual Symbolism and Aesthetic Strategies of 3D Graphic-Based Movie Posters
Abstract : This study analyzes the visual symbolism and aesthetic strategies of 3D graphic-based movie posters through the lens of Rudolf Arnheim’s theory of visual perception. The analysis focuses on the official posters of Avatar: The Way of Water, Dune: Part Two, and Doctor Strange in the Multiverse of Madness, examining them in terms of formal elements, symbolic meaning, and perceptual guidance. Avatar visualizes the themes of family and water through a vertical layout and strong color contrast, while Dune conveys its desert setting and narrative gravity through a central axis composition and dominant orange tones. Doctor Strange represents the chaos of the multiverse using a radial layout and fragmented imagery. In all three cases, 3D graphics enhance spatial depth and dimensionality, effectively drawing the viewer’s gaze inward and conveying each film’s atmosphere and worldview. Central placement, symbolic imagery, and dynamic contrasts function as key strategies in guiding visual perception and sustaining attention. These findings confirm that contemporary movie posters are not merely promotional tools but immersive visual media and independent artistic forms with aesthetic and cultural significance.
Keyword : Movie Poster, Visual Symbolism, Aesthetic Strategies, Visual Perception Theory
http://dx.doi.org/10.29056/jncist.2025.12.06
Abstract : The purpose of this study was to investigate the effect of aesthetic factors in cosmetic package design on female consumers' purchasing behavior against the backof the rapid growth of the Chinese cosmetics market and increasing attention to domestic brands. In particular, the study focused on highly portable product categories and analyzed how consumers' aesthetic desires and conspicuous desires affect purchase intentions through package design. The study employed a two-stage survey of Chinese female consumers aged 19 to 50. In the first survey, a cross-tabulation analysis was conducted based on age and occupation to identify cosmetic product categories that are carried when going out. The product categories were divided into three groups kin care, makeup, and perfume and cream, cushion foundation, and portable perfume cases, which showed the highest carrying frequencies within each category, were selected as the focal products of the study. In the second survey, the effects of visual elements in package design on consumers' purchasing behavior were analyzed. The results confirmed that all five visual elements constituting the package shape, color, illustration, collaboration, and brand logo had a positive effect on consumers' aesthetic desires. In particular, collaboration and brand logo were found to be key factors that stimulate consumers' desire for conspicuous display. In addition, a majority of respondents indicated that package design directly affects purchasing behavior, and it was empirically demonstrated that aesthetics play an important role in actual purchasing decisions for highly portable cosmetic products.
Keyword : cosmetic package design, aesthetics, portability, consumer purchasing behavior
http://dx.doi.org/10.29056/jncist.2025.12.07
Abstract : The rapid evolution of the digital information environment has led to a major paradigm shift in knowledge organization within library and information science. The emergence of conceptual models such as FRBR and the IFLA Library Reference Model (LRM), together with the transitionthe existing cataloging rules, AACR to RDA, and in Korea, the revision of KCR4 to KCR5, brought about a paradigm shift in cataloging practices and data structures overall. This study explores library and information science students' experiences in cataloging education and identifies effective teaching strategies for this transitional era. Findings indicate that although students value cataloging for academic and career development, they face challengesunfamiliar terminology, complex rules, and limited class time. Accordingly, this study proposed the following approaches: concept-principle-centered understanding, pre- and post-rule comparative learning, expanded case-based and participatory practice, and learner-centered instructional design. These strategies aim to enhance cataloging education and support the development of future-oriented curricula in knowledge organization.
Keyword : knowledge organization education, cataloging instruction, RDA, KCR5, teaching strategies, LIS education
http://dx.doi.org/10.29056/jncist.2025.12.08
Research Trends and Future Challenges of Digital Education for Young Children
Abstract : Over the past decade, research on digital education for young children in South Korea has concentrated on four major research trends: the developmental effects of smart media exposure, prevention and intervention strategies for media overdependence, software education to cultivate computational thinking, and digital citizenship and digital literacy education. Previous studies have identified the significant influence of smart device usage patterns-specifically timing and frequency-on children's developmental levels, empathy capacity, and self-regulation abilities. Moreover, systematic parental mediation has been demonstrated to play a positive role in supporting healthy child development. As smart media overdependence has emerged as a critical social concern, both preventive education and systematic intervention strategies implemented by parents and teachers have proven effective in enhancing children's self-control and reducing media dependency. Software education centered on computational thinking has been shown to simultaneously advance problem-solving abilities, creativity, and scientific-technological interest through play-based and experiential learning approaches, while also demonstrating potential for interdisciplinary curriculum integration. Digital citizenship education, focusing on media literacy and the cultivation of responsible digital citizens, has emerged as essential for young children preparing to participate in digital society. Based on these research trends, future research should prioritize the development of developmentally appropriate digital education programs linked to play-centered curricula, the strengthening of digital competencies among parents and teachers, and the establishment of measures to bridge digital divides and ensure safe digital environments for all children.
Keyword : Early Childhood Digital Education, Smart Media, Digital Literacy, Research Trends
http://dx.doi.org/10.29056/jncist.2025.12.09
The Effects of College Students' Adaptation Capacity and Stress on Academic Achievement
Abstract : This study examined how adaptive capacity and stress factors affect academic achievement among college students. A total of 51 freshmenthe humanities program at a private college participated in the student adaptation capacity assessment. The results showed that positive psychological factors such as future adaptability, goal orientation, and self-esteem were significantly associated with higher academic achievement. Among these, future adaptability was most strongly linked to better academic outcomes. In contrast, self-control, current adaptability, and familysocial conflicts were not directly related to academic performance. Regression analysis identified current adaptability and academic burden as significant predictors of academic achievement, with academic burden identified as a major risk factor for lower performance. The findings help inform the design of college programs that reduce academic and career-related stress. Furthermore, the results suggest the need for targeted support focusing on present psychological adaptation and stress management to enhance students' academic success.
Keyword : College Students' Adaptation Capacity, Adaption, Stress, Academic Burden, Academic Achievement
http://dx.doi.org/10.29056/jncist.2025.12.10
Factors Affecting Participation in Professional Volunteer Activities among Older Adults
Abstract : This study aims to identify the factors influencing elderly participation in professional volunteer activities. For this purpose, datathe 2023 Social Survey conducted by Statistics Korea were analyzed, focusing on 716 individuals aged 65 and older who reported participating in volunteer activities. Analytical methods included frequency and descriptive statistics, T-tests, Chi-square tests, and logistic regression analysis. The influencing factors were categorized into socio-demographic factors, health-related factors, and social factors. The T-test results revealed statistically significant differences between those who engaged in professional volunteer activities and those who did not, in terms of age, education level, perceived social class, and social activity involvement. Logistic regression analysis indicated that age, education level, and social activity involvement significantly influenced elderly participation in professional volunteer activities. More specifically, the results indicated that as age increases, the likelihood of older adults participating in professional volunteer activities decreases and a higher level of education was associated with a greater likelihood of participation. Additionally, older adults who were more actively engaged in social activities were more likely to participate in professional volunteer activities. The implications of these findings were discussed.
Keyword : Older adults, Professional volunteer activities, Human capital, Logistic regression analysis
http://dx.doi.org/10.29056/jncist.2025.12.11
A Qualitative Study on AI-Based Education and Integrated Support Systems for Career-Interrupted You
Abstract : This study aims to explore the challenges experienced by career-disrupted youth and their perceptions of AI technology education, and to propose educational strategies that support their career restoration. To this end, in-depth interviews were conducted using a qualitative research approach with expertsyouth support organizations and young adults who had experienced career interruption. The analysis revealed several key findings. First, career-disrupted youth faced multifaceted difficulties such as repeated job-search failures, low self-efficacy, social isolation, and limited access to information, and expressed the need for practical support and structured roadmaps for career redesign. Second, participants recognized AI technology as an essential job competency and demonstrated strong interest and willingness to participate in practice-oriented AI education such as AI coding, graphic design, and cloud fundamentals. Third, while participants acknowledged the flexibility of online learning in terms of time and location, they also noted limitations such as insufficient immediate feedback and the burden of self-directed learning, which led them to prefer blended formats, hands-on practice, and portfolio-based instruction. Fourth, essential support elements for AI-based learning included mentoring, AI learning tutors, employment linkage programs, and financial assistance to sustain academic engagement and career development. This study provides foundational insights for designing AI-enhanced educational and support systems grounded in the actual contexts of career-disrupted youth and suggests that strengthening AI-based competencies can facilitate their social reintegration and contribute to national human resource development.
Keyword : career-interrupted youth, career disruption experience, educational needs, practical skill enhancement, AI technology education, career support, qualitative research
http://dx.doi.org/10.29056/jncist.2025.12.12
Table of Contents