Jurnal Media Computer Science

Data Mining, Decision Tree, C4.5, Community Satisfaction Index

Doan Sefriyansyah — 2026-07-22

This study aims to apply a data mining method based on the C4.5 Decision Tree algorithm to classify the Public Satisfaction Index (PSI) regarding services at the Central Aceh District Attorney’s Office. Until now, the data from the CSI questionnaires collected has not been optimally utilized because it has only been processed using average calculations, thus failing to provide in-depth information regarding patterns of public satisfaction. The C4.5 Decision Tree method was used because it is capable of producing a classification model that is easy to understand and can identify the service attributes that most influence the level of public satisfaction. The data used in this study were derived from questionnaire results based on seven criteria: service procedures, service information, processing time, staff competence, technological facilities, complaint handling, and integrity. The results indicate that the C4.5 algorithm can be effectively applied to classify levels of public satisfaction and generate decision trees that serve as a basis for evaluation and decision-making.

Evaluasi Tingkat Kematangan Tata Kelola Ti Di Pos Kediri Menggunakan Cobit 5

Ahmad Muharram Alfarisi — 2026-07-22

Kediri Post Office, as a public organization that provides vital communication and financial services to the community, requires well-developed and effective information technology (IT) infrastructure and management. However, no comprehensive evaluation of the maturity level of IT governance has previously been conducted to identify the current condition and potential areas for improvement. This study aims to evaluate the maturity level of IT governance at Kediri Post Office using the COBIT 5 framework across three main domains (APO, DSS, and MEA), with a focus on identifying gaps and providing recommendations for process improvement. The research employed a qualitative approach, with data collected through in-depth interviews, field observations, and document analysis to validate the implementation of IT processes within the organization. The evaluation results indicate that the overall IT governance maturity level at Kediri Post Office is at Level 2 (Repeatable Process), with an average score of 2.33 on a scale of 0–5. Two processes have achieved Level 3 (Established Process), namely APO03 (Manage Enterprise Architecture) and DSS02 (Manage Service Requests and Incidents), while the remaining seven processes are still at Level 2. Gap analysis revealed an average gap of 0.89 levels compared to the target maturity level of 3.33. The primary issues identified include incomplete formal documentation, inconsistent process formalization across organizational units, and monitoring and evaluation mechanisms that are not yet systematic, particularly within the MEA (Monitor, Evaluate, and Assess) domain.

Prediksi Harga Kos di Wilayah Jimbaran Menggunakan Algoritma Decision Tree Regression Berbasis Web

I Putu Krishna Paradipta — 2026-07-22

Jimbaran District is one of the strategic areas in Badung Regency, Bali, experiencing rapid growth in the education, tourism, and economic sectors. This condition has increased the demand for temporary housing, especially boarding houses for students and migrant workers. The large number of boarding house options with various facilities and rental prices often makes it difficult for prospective tenants to choose accommodations that match their needs and financial capabilities. This study aims to develop a web-based boarding house price prediction application using the Decision Tree Regression algorithm to assist users in estimating boarding house prices based on the desired facilities. The research applies a quantitative approach using machine learning and data mining methods. The dataset consists of boarding house data in the Jimbaran area, including room size, bathroom type, room facilities, public facilities, distance to campus, and monthly rental price. The research stages include data preprocessing, feature transformation, dataset splitting using the Train-Test Split method, model training, and model evaluation using MAE, RMSE, and R² metrics. The results show that the Decision Tree Regression algorithm achieved good performance with an R² value of 0.8332, an MAE value of IDR 225,462, and an RMSE value of IDR 302,509.

Design of Inventory System at Putri Gemilang Crackers Factory Using End-User Development Method

Dela Nabila Apriyani — 2026-07-22

This study aims to design a web-based inventory information system at Pabrik Kerupuk Putri Gemilang using the End-User Development (EUD) method. The problems found in the inventory management process were that it was still carried out manually using notebooks, causing recording errors, delays in report preparation, and discrepancies between stock records and physical stock in the warehouse. The research method used was a qualitative method with data collection techniques through observation and interviews. The End-User Development approach was applied because it involves users directly in the process of system requirement analysis and system design, so the resulting system is more suitable for operational needs. The results of the study indicate that the web-based inventory system is able to assist the management of item data, incoming goods, outgoing goods, and inventory reports more effectively and efficiently. The system is also able to display real-time stock information and generate automatic reports to support monitoring and decision-making processes. In addition, the system was designed with a simple and user-friendly interface to make it easier for users to operate the system. The implementation of this web-based inventory information system is expected to improve the effectiveness, accuracy, and integration of inventory management at Pabrik Kerupuk Putri Gemilang.

The Implementation Of K-Means Clustering Method To Product Sales Data At Pratama Elektronik Shop, Taba Lagan Village, Central Bengkulu Regency

Yoga Putra Pratama — 2026-07-22

The implementation of K-Means Clustering method to product sales data at Pratama Elektronik shop, Taba Lagan Village, Central Bengkulu Regency, can assist store management in grouping sales data automatically and in a computerized manner, as well as providing information regarding product sales patterns that can be used to support decision-making, particularly in inventory management. The clustering results consisted of three groups: Cluster C1 (high-sales category), Cluster C2 (medium-sales category), and Cluster C3 (low-sales category). Based on testing using 100 sales data records from Pratama Elektronik shop, Taba Lagan Village, the results showed that 76 products were classified in the low-sales category, 21 products were classified in the medium-sales category, and only 5 products were classified in the high-sales category.

Comparative Analysis Of Fuzzy Time Series And Double Exponential Smoothing Methods In Predicting Medicine Demand At Community Health Centers

Aldo Zantohodi — 2026-07-22

Drug inventory management in primary health centers requires accurate forecasting to reduce stock-outs, overstock, and logistics inefficiency. This study compares Cheng Fuzzy Time Series (FTS) and Holt Double Exponential Smoothing (DES) for forecasting drug demand at Pajar Bulan Public Health Center. Using LPLPO data, 84 valid drug items were analyzed with January-December 2025 as training data and January-April 2026 as testing data. DES produced a lower MAE, but FTS achieved a better MAPE and was superior on 65.5% of items with valid MAPE. The results indicate that FTS is more robust for fluctuating active drug demand, while DES is suitable for more stable trend patterns.

Application Of The Term Frequency-Inverse Document Frequency (TF-IDF)-Based Support Vector Machine (SVM) Method For Sentiment Classification Of Customer Reviews On My Lova Bengkulu

Muhammad Sihab — 2026-07-22

This study aims to apply a Term Frequency-Inverse Document Frequency (TF-IDF)-based Support Vector Machine (SVM) method for sentiment classification of My Lova Bengkulu customer reviews. The research data consisted of 109 reviews obtained from Google Reviews, which were then subjected to a preprocessing process involving cleaning and stemming. Next, weighting was performed using TF-IDF, and classification was carried out using the SVM algorithm. The results showed that 93 reviews (85.32%) were positive, while 16 reviews (14.68%) were negative. The model achieved an Accuracy of 81.82%, Precision of 81.80%, Recall of 100%, and an F1-Score of 90.00%. These results demonstrate that the TF-IDF-based SVM method is capable of effectively classifying customer sentiment.

Classification Of Papaya Leaf Diseases Using The Convolutional Neural Network Method With The Mobilenetv3 Architecture

Komang Bagus Sadewa Netra — 2026-07-23

This study aims to develop a classification system for papaya leaf diseases based on digital image processing using the Convolutional Neural Network (CNN) method with the MobileNetV3 architecture. The background of this research is the manual process of identifying papaya leaf diseases by farmers, which is often inefficient and prone to misdiagnosis. The research method adopts a software engineering approach using the Agile development model, allowing iterative and flexible system development. The dataset consists of papaya leaf images categorized into three classes: curl, ringspot, and healthy, obtained from field observations and secondary datasets. The data were processed through preprocessing stages before being used to train the CNN model. The results indicate that the best model was achieved using a learning rate of 0.001 and 30 epochs, with a validation accuracy of 95.56% and a testing accuracy of 88.89%. The model demonstrates high confidence in classifying images, particularly for the curl and healthy classes. However, the confusion matrix reveals that the model's performance on the ringspot class remains relatively low due to a high misclassification rate. Overall, the developed system is capable of automatically identifying papaya leaf diseases and has strong potential for implementation as an Android-based application to support early detection and decision-making in plant care.

Design Of A Clothing Sewing Service System At Bucik Rendo Tailor User Prototype Method

Bima Herbudi — 2026-07-23

This study aims to design a clothing sewing service information system at Bucik Rendo Tailor using the prototype method. The problem that occurs in this sewing service business is the data management process that is still done manually, such as recording customer data, orders, payment transactions, and business reports. This condition causes frequent recording errors, difficulties in data retrieval, and slow service to customers. To overcome these problems, this study uses the prototype method because it can help developers understand user needs more clearly through a gradual evaluation and improvement process. Data collection was carried out through observation, interviews, literature studies, and documentation studies. The results of the study are in the form of a system design that includes a Data Flow Diagram (DFD), Entity Relationship Diagram (ERD), database design, and user interface design. The designed system has features for managing customer data, products, orders, transactions, work status, and reports. With this system design, it is expected to help improve the effectiveness of clothing sewing service business management, facilitate service to customers, and serve as a reference in the development of a web-based sewing service information system in the next implementation stage.

Usability Evaluation And User Acceptance Testing (Uat) On Website-Based Digital Psychology Testing Application: A Case Study Of Psikotestyuk.Id

Adi Putra Dwi Nugraha — 2026-07-23

Digital transformation in the recruitment process through digital psychological testing applications such as psikotestyuk.id aims to increase the efficiency and objectivity of candidate selection. However, the effectiveness of this system is highly dependent on the ease of use and functional acceptance by end users, both prospective employees and HR. This study aims to evaluate the level of usability and user acceptance of the psikotestyuk.id application to identify operational obstacles and areas for system architecture improvement. The method used is Mixed Methods, integrating quantitative data from Usability Testing on 30 prospective employee respondents, and qualitative data from User Acceptance Testing (UAT) with in-depth interviews with three HR staff. The results showed an average usability score in the "Very Good" category with a value range of 4.21 to 4.40. Although functionally the system was very well received, UAT testing revealed the need for server infrastructure optimization to handle high traffic and the addition of a bulk data import feature to improve the efficiency of mass recruitment administration. In conclusion, the psikotestyuk.id application is very suitable for continued use with several development notes on the data scalability side.

Payment System Service Quality Analysis Qris Uses The Servqual Method at The Café Trilogi

Pemi Afely Athin — 2026-07-23

The development of digital payment systems has encouraged culinary business actors to adopt the Quick Response Code Indonesian Standard (QRIS) as an alternative to faster, practical, and more efficient transactions. However, the success of QRIS implementation is not only determined by its level of use, but also by the quality of service perceived by customers. This study aims to analyze the service quality of the QRIS payment system at Cafe Trilogi based on the gap between customer perception and expectations using the Service Quality (SERVQUAL) method. The study used a descriptive quantitative approach involving 100 respondents selected through purposive sampling techniques, namely Cafe Trilogi customers who had made transactions using QRIS. Data was collected through the distribution of questionnaires that had met the validity and reliability tests, then analyzed using SERVQUAL Gap Analysis on five dimensions of service quality, namely tangible, reliability, responsiveness, assurance, and empathy. The results showed that the quality of QRIS services at Cafe Trilogi obtained a positive gap value of +1.42, indicating that the services provided have met customer expectations. The assurance dimension had the highest gap value (+0.72), while the reliability dimension showed a negative gap (-0.01), indicating room for improvement in service consistency and network stability.

Implementasi Automated Conflict Testing Untuk Mencegah Double Booking Pada Sistem Penjadwalan Klinik Gigi

Sulthan Zahran Sunata — 2026-07-23

Double booking poses a significant risk to dental clinic scheduling systems, as it disrupts service workflows, increases patient wait times, and reduces practitioner operational efficiency. While healthcare scheduling literature extensively covers slot optimization, resource allocation, and no-show issues, there is limited research demonstrating how conflict prevention rules are automatically validated within implemented clinic systems. Conversely, software testing literature tends to focus on general automation rather than executable testing for rule-based conflict checking. This study aims to implement automated conflict testing to verify double-booking prevention mechanisms in a Django REST API-based dental clinic scheduling system. The methodology involves automated testing at both the model and API levels, utilizing scenarios derived from conflict-checking business rules. Test results indicate that all scenarios passed and that conflict-checking rules were consistently applied to both initial booking and rescheduling processes. The study's primary contribution is an executable, structured automated conflict-testing approach that confirms the acceptance of back-to-back slots, the rejection of active overlaps, and the subjection of rescheduling requests to conflict validation.

Densenet201 Feature Extraction With Soft Voting Ensemble For Accurate Rice Leaf Disease Classification

Nelly Khairani Daulay — 2026-07-23

Rice leaf diseases are one of the major factors contributing to reduced agricultural productivity and economic losses for farmers. Manual disease identification generally requires expert knowledge and is often difficult to perform efficiently in field conditions. Therefore, this study aims to develop a rice leaf disease classification system by combining DenseNet201 as a feature extractor and a Voting Ensemble approach as the classifier. The dataset consisted of 1,470 rice leaf images categorized into five classes: Bacterial Leaf Blight, Brown Spot, Healthy Leaf, Leaf Blast, and Tungro. The dataset was divided using a stratified split strategy into 80% training data, 10% validation data, and 10% testing data. Image augmentation was applied only to the training set, increasing the number of training samples to 7,056 images. DenseNet201 was employed to extract image features into 1,920-dimensional feature vectors, which were subsequently classified using Logistic Regression, Support Vector Machine (SVM), Hard Voting, and Soft Voting. Experimental results showed that Logistic Regression achieved an accuracy of 95.24%, while SVM achieved 95.92%. Hard Voting obtained an accuracy of 95.24%, whereas Soft Voting achieved the best performance with an accuracy of 95.92%, precision of 95.75%, recall of 95.70%, F1-score of 95.71%, and ROC-AUC of 99.76%. Furthermore, the best-performing model was deployed in a Streamlit-based application for automatic rice leaf disease identification. The results demonstrate that the combination of DenseNet201 and Soft Voting provides an accurate and effective approach for rice leaf disease classification and has strong potential as an early disease detection tool in agriculture.

Does CFTC Regulation Reduce Prediction Market Anomalies? A Benford’s Law And DiD Analysis

Al Berlant Ghulam Fania Akhmad — 2026-07-23

The rapid growth of prediction markets raises concerns about data integrity and susceptibility to manipulation. This study examines whether Commodity Futures Trading Commission (CFTC) regulation of Polymarket reduces market anomalies measured through Benford's Law conformity. Employing a quasi-experimental nonequivalent control group design with a difference-in-differences (DiD) estimator, the study exploits the CFTC Amended Order of 25 November 2025 as the treatment on Polymarket, with Kalshi (regulated since 2020) as the control group. Daily price and volume data for both platforms were retrieved from Dune Analytics for January 2023–June 2026 (over 1.4 billion observations) and transformed into monthly Mean Absolute Deviation (MAD) per platform per category; the volatility index (VIX) serves as a covariate. DiD estimates on the theoretically valid volume outcome reveal no significant regulatory effect (β = +0.0038; p = 0.073); thus the hypothesis that regulation reduces anomalies is not supported, as Benford conformity was already high and improved on both platforms due to market maturation. Placebo and pre-trends robustness tests confirm the validity of the design. This study constitutes the first Benford–DiD quasi-experimental test in prediction markets.

A Comparative Analysis of User Sentiment and Review Patterns on The Dana and Seabank Apps Using The Naive Bayes, Support Vector Machine and Logistic Regression Algorithms

Nur Ayu Siti Hardianti — 2026-07-23

The development of financial technology (fintech) in Indonesia has driven a significant increase in the use of digital financial applications, particularly e-wallets and digital banks. The DANA application as a representative of e-wallet services and SeaBank as a representative of digital banking services have attracted millions of users with different service characteristics. These differences in service characteristics potentially generate different patterns of user sentiment and reviews. This study aims to analyze and compare user sentiment and review patterns of the DANA and SeaBank applications using three machine learning algorithms: Naive Bayes, Support Vector Machine (SVM), and Logistic Regression. User review data were collected from the Google Play Store and processed through text preprocessing stages including case folding, tokenization, stopword removal, and stemming. Text features were extracted using the TF-IDF method before being classified into positive, negative, and neutral sentiment categories. Model performance evaluation was conducted using accuracy, precision, recall, and F1-score metrics. The comparative analysis results are expected to provide a comprehensive overview of differences in user sentiment based on the type of fintech service, while also identifying the most optimal machine learning algorithm for sentiment analysis of digital financial application reviews.

Implementasi Data Mining Algoritma K-Nearest Neighbor Pada Penentuan Pemberian Sanksi Disiplin Di Rumah Sakit Umum Daerah Dr. M. Yunus Bengkulu

Fiqi Aprianto — 2026-07-23

Regional General Hospital (RSUD) Dr. M. Yunus Bengkulu is a government-owned health care facility with a strategic role in providing health services to the public, making employee discipline an important factor to maintain. The determination of disciplinary sanctions for employees at RSUD Dr. M. Yunus Bengkulu is still carried out manually based on recorded violation data, which requires a relatively long time and has the potential to create subjectivity in decision making. This study aims to apply the data mining method using the K-Nearest Neighbor (K-NN) algorithm to classify employee disciplinary sanctions based on five assessment attributes, namely years of service, number of tardiness incidents, number of unexplained absences, SOP violations, and service ethics violations. The system was developed using the Waterfall development method and implemented as a web-based application using the PHP programming language and MySQL database. Testing was conducted on 20 training data and 20 testing data using a K value of 5, calculated from the square root of the number of training data. The test results show that the K-Nearest Neighbor algorithm is able to classify employee disciplinary violation data into four sanction categories, namely no sanction, verbal warning, written warning, and disciplinary sanction, with an accuracy rate of 90%. Black box testing results also show that all system functions run according to requirements. This research is expected to help the management of RSUD Dr. M. Yunus Bengkulu make decisions on employee disciplinary sanctions in a more objective, consistent, and measurable manner.