Journal of Digital Business and Data Science

Design and Development of an Android-Based POS Application for MSME Sales Optimization

Bagus Karandika — 2026-06-09

Backround: The rapid growth of Micro, Small, and Medium Enterprises (MSMEs) in Indonesia has created an urgent demand for affordable yet robust digital management tools. Conventional cash register systems frequently fail to deliver real-time inventory control, structured sales reporting, or multi-role access management constraints that impede operational efficiency and business scalability.

Objective: This study aims to design and develop a hybrid Android-based Point of Sale (POS) application, designated GS Baby Care POS, specifically engineered to address those operational gaps in the MSME retail sector.
Method: The application was built using an agile development methodology combining Vue.js 3, Capacitor JS, and a Supabase (PostgreSQL) cloud backend, delivering a JAMStack-plus-Mobile-Hybrid architecture.
Findings and Implications: System evaluation employed Black-Box Functional Testing and the System Usability Scale (SUS) questionnaire administered to fifteen respondents. Functional testing confirmed that all core modules including role-based authentication, variant-level inventory management, shift tracking, customer relationship management, and financial dashboards — operated without defect. SUS assessment yielded a mean score of 83.7 (Grade B, Excellent), indicating high user acceptance. The application demonstrably reduced average transaction processing time by 42 percent and provided real-time business analytics previously unavailable to the target enterprise.
Conclusion: These results confirm that a cloud-native, hybrid-mobile POS system constitutes a viable and cost-effective solution for MSME digital transformation.

Implementation of Prototyping in AI-Based Job Portal Development to Improve User Experience

Evelin Salsabila Aflahah — 2026-06-12

The rapid development of digital technology has significantly changed the recruitment system, particularly through the emergence of AI-based job portals. However, many systems still fail to meet user expectations due to a lack of user involvement early on in the development process. This study aims to analyze the implementation of prototyping in the development of AI-based job portals to improve user experience and system effectiveness. This research adopts a design-based research approach, using user needs analysis, prototype development, and iterative evaluation. Prototyping is carried out through several stages, including low-fidelity and high-fidelity models, to ensure the usability and functionality of the system. The findings show that prototyping allows for early validation, reduces development risk, and improves user satisfaction through continuous feedback integration. Additionally, artificial intelligence integration supports personalization and improves job matching accuracy. The study concludes that prototyping plays an important role in developing a user-centric digital platform and significantly contributes to improving the quality of the system and user experience. Further research is recommended to explore AI-based prototyping for more adaptive and intelligent system design.

Implementation of Data Mining for Customer Segmentation Using the K Means Clustering Algorithm Based on Annual Income and Spending Score

Fortina Lumban Gaol — 2026-03-07

Background: This research is motivated by the dynamics of the retail industry, which requires a deep understanding of consumer behavior in order to compete effectively in an increasingly competitive market. Many marketing strategies fail to achieve optimal results because they overlook variations in individual shopping behavior within large customer populations. Understanding these behavioral differences is important for developing more targeted and effective marketing strategies.

Objective: This study aims to group customers into homogeneous segments in order to support more precise strategic decision-making in marketing activities.

Method: The study applies a data mining approach using the K-Means clustering algorithm to analyze a dataset consisting of 200 customers. The clustering process is conducted based on two main variables, namely annual income and spending score, to identify patterns of consumer behavior.

Findings and Implications: The results reveal five distinct consumer clusters with different behavioral characteristics. The Target group represents the majority with 81 customers, followed by the Sultan group (39 customers), the Thrifty group (35 customers), the Passive group (23 customers), and the Impulsive group (22 customers). The findings indicate that income level does not always correlate linearly with consumption intensity, implying that behavioral-based segmentation provides more accurate insights for marketing strategy development.

Conclusion: Customer segmentation using the K-Means clustering algorithm enables clearer identification of target markets through well-defined cluster separation. Therefore, marketing strategies should emphasize lifestyle orientation rather than focusing solely on purchasing power to optimize customer loyalty and engagement.

Diagnostic Expert System Website-Based Stroke Disease Using Forward Chaining and Certainty Factor Methods

Muhammad Fikri Bagus Pratama — 2026-06-09

Background: Stroke is a neurological condition characterized by the sudden loss of brain function resulting from disruption of blood supply to the brain. It ranks as the second leading cause of death globally, with a mortality rate ranging from 18% to 37%, and constitutes a major cause of neurological disability in Indonesia as well as the third leading cause of death worldwide.
Objective: This study aimed to develop a web-based expert system enabling patients and their families to perform early detection of stroke symptoms.
Method: This study employed a prototype-based development methodology. The knowledge base was constructed through structured interviews with a neurologist and validated through cross-checking with clinical records. The Forward Chaining method served as the inference engine, deriving diagnostic conclusions from symptom-based facts, while the Certainty Factor method quantified diagnostic uncertainty. System testing was conducted using six patient case samples provided by the expert.
Findings and Implications: The system achieved a diagnostic accuracy of 86.68% based on cross-validation with expert knowledge using six clinical case samples. Black-box functional testing confirmed that all system features performed as expected.
Conclusion: These results indicate that the system is capable of supporting preliminary stroke symptom assessment, thereby facilitating early decision-making prior to professional medical consultation. However, given the limited number of test cases, the system’s generalizability warrants further validation using a larger clinical dataset.