Bulletin of Computer Science Research

Perancangan Sistem Arsip Data Digital Menggunakan Model Waterfall Berbasis Web

2025-07-28T09:34:22+07:00

This research aims to design a digital data archiving system at PT Perkebunan Nusantara Regional I Kebun Cisalak Baru. This system was developed to replace the existing manual process, making archive management faster, easier, and more secure. The system was built using PHP Native and MySQL, and tested using the black box method to ensure its functionality runs smoothly. The development results show that the system is able to manage archive data effectively, from input and search to report printing. With this system, administrative processes become more efficient and organized.

Sistem Pendukung Keputusan Pemilihan Perangkat Umum Persmabimed Berbasis Web dengan Metode Profile Matching

2025-06-03T17:35:07+07:00

Persmabimed is an organization under Universitas Negeri Medan that serves as a platform for students receiving the KIP-K scholarship and plays an important role in managing member activities and information. However, the selection process for general board members such as the chairman, secretary, and treasurer is still conducted manually and tends to be subjective, potentially leading to a mismatch between candidates and their assigned roles. This study aims to design and implement a web-based decision support system using the Profile Matching method to assist in the selection process of Persmabimed board members. The method is chosen for its ability to compare candidate profiles with ideal profiles based on GAP values and criterion weights. Data was collected through observation, interviews with organization administrators, and literature studies. The selection process involves calculating core and secondary factors, followed by ranking based on a weighted combination of hard and soft skills. The system’s results demonstrate that the Profile Matching method can produce objective and accurate decisions, selecting Irvan Affandi as chairman, Dinda Rizky Fadilah as secretary, and Kiki Ratna Sari as treasurer. The system was developed using PHP and Bootstrap to ensure accessibility and streamline the selection process. This research improves efficiency and fairness in organizational decision-making and can be applied to similar organizations in the future.

Penerapan Model Least Square Untuk Prediksi Nilai Dalam Sistem Pengabdian Berbasis Website

2025-07-26T16:03:31+07:00

This study aims to develop a web-based internship information system (PKL) at SMK At-Taqwa Muhammadiyah Miri to replace the manual process of recording student attendance and daily activity reports. The main problem addressed in this research is the inefficiency of PKL administrative processes, which are still carried out manually, leading to a high risk of data loss, delayed reporting, and low accuracy in student performance evaluation. In addition, monitoring student activities by supervisors and teachers is less than optimal due to the absence of a real-time integrated reporting system. The system was built using the Waterfall method with the Laravel framework and MySQL database, and it integrates a Least Square regression model to predict students’ final scores based on their attendance and daily activities. The system testing results showed 100% functional success across 16 test scenarios using the Black Box method. This system is expected to improve efficiency, transparency, and objectivity in monitoring and evaluating internship activities.

Penggunaan Metode Naïve Bayes untuk Klasifikasi Topik Tweet Bidang dan Non-Bidang Rektorat Telkom University pada Akun Telyufess

2025-07-19T23:34:41+07:00

Student complaints submitted through the Telyufess account on the social media platform X have not been optimally utilized as input for evaluating campus services at Telkom University. This study aims to classify tweets from the Telyufess account into two categories: domain-related (linked to official university units such as academics, finance, and campus services) and non-domain (general complaints unrelated to specific units). The main issue addressed is the need for an automated mapping system of student complaints to support campus service evaluations. The classification method used is Naïve Bayes, involving manual labeling by the researcher and assistant annotators (with inter-rater validation), text preprocessing (normalization using a standard dictionary and the Sastrawi library, removal of special characters, stop word filtering based on Indonesian language lists augmented with the unique term “telyu!”), tokenization, stemming, TF-IDF weighting, and dataset splitting in ratios of 65:35, 70:30, 80:20, and 90:10. A total of 1,090 tweets were collected between January 1, 2023 and January 1, 2025 using the Tweet Harvest API, based on criteria including complaints, opinions, and suggestions (retweets were excluded). The highest accuracy was achieved at 87.27% with a 90:10 split, followed by 81.74% (80:20), 78.35% (70:30), and 76.96% (65:35). Although the model showed signs of overfitting on training data (accuracy >99%), the results demonstrate that Naïve Bayes is effective for automated tweet classification and contributes to the use of social media as a data source for evaluating campus services.

Sistem Informasi Keamanan Kunci Pintu Pintar Berbasis Internet of Things Notifikasi Real-Time Berbasis WhatsApp

2025-07-31T14:11:33+07:00

Security remains a primary concern in domestic and marketable structures, especially as conventional door cinches come decreasingly vulnerable to duplication and forced access. This study presents the design and perpetration of a smart door lock security system grounded on Internet of Things and Radio frequence Identification technologies. The main ideal is to give a real- time, ever controlled locking medium integrated with a mobile- grounded messaging platform. The development process adopts the nimble methodology, which promotes inflexibility, responsiveness to stoner feedback, and rapid-fire replication in software and tackle integration. The tackle armature utilizes ESP32 and Arduino Nano microcontrollers to manage essential factors, including RFID compendiums, relays, buzzers, solenoid cinches, LED and LCD as display module. On the software side, the system incorporates a pall connected messaging operation to deliver real- time announcements to druggies, including cautions for unauthorized access attempts, successful entries, and low battery warnings. Data communication between the microcontrollers is enforced via periodical protocol, while the commerce with the messaging service is established using the HyperText Transfer Protocol. Testing involved colorful scripts including valid access, unrecorded card attempts, and remote commands. The testing scenario uses three RFID cards and a WhatsApp chatbot, with one card as the master and the other two as access cards, with each card undergoing three trials. Results demonstrate a high success rate in card discovery and command prosecution with an average system response time below one second. The system also allows druggies to register and remove access cards via a master card medium, which enhances usability and control. This exploration contributes a low- cost, customizable, and effective security result suitable for smart homes, boarding houses, and small- scale marketable parcels. It shows the feasibility of integrating open- source microcontrollers with everyday messaging tools to achieve effective, real- time home security. The proposed system provides a foundation for unborn exploration in combining internet grounded control systems with biometric detectors or machine literacy grounded access analytics such as developing a separate mobile application so that it does not use third-party applications, allowing users to obtain more accurate feedback through door leaf detector sensors.

Analisis Performa Variational Quantum Classifier (VQC) dengan ZZ Feature Map dan Angle Encoding Untuk Mengidentifikasi Serangan Jantung

2025-07-24T00:06:15+07:00

Heart attact is one of the leading causes of death worldwide, with mortality increasing due to delayed diagnosis and limited facilities in some regions. Early detection during the crucial first hour after symptoms appear (the golden hour) is crucial for reducing mortality and improving patient prognosis. This study aims to evaluate the performance of the Quantum Neural Network (QNN) by implementing the Variational Quantum Classifier (VQC) model using two types of feature maps: ZZ Feature Map and Angle Encoding, for heart attack detection classification using datasets from Kaggle. The research process includes dataset collection, Exploratory Data Analysis (EDA), data preprocessing and splitting, model building using ZZ Feature Map and Angle Encoding, ending with model performance evaluation. The results showed that VQC using ZZ Feature Map achieved an accuracy of 52.27% with a confusion matrix showing suboptimal predictions and relatively low precision, recall, and F1-score values. Meanwhile, VQC using Angle Encoding achieves an accuracy of 68.18% with a confusion matrix that shows a higher number of correct predictions and better precision, recall, and F1-score results compared to VQC using ZZ Feature Map. Evaluation using a confusion matrix shows that the model with Angle Encoding can minimize the prediction error. However, the accuracy achieved is not yet optimal for direct clinical implementation. These findings underscore the need for further development to enhance model performance, while serving as a promising first step toward establishing a foundation for developing more optimal QNN methods in the future.

Integrasi PIPRECIA dan MACROS Dalam Penentuan Lokasi Pertanian Berkelanjutan

2025-07-31T21:20:54+07:00

This study develops a decision support system for selecting sustainable agricultural locations amid the challenges of climate change. Two integrated multi-criteria decision-making (MCDM) methods are employed: PIPRECIA (Pivot Pairwise Relative Criteria Importance Assessment) to determine the importance weights of criteria based on expert rankings, and MACROS (Measurement Alternatives and Ranking according to the Compromise Solution), a structured method used to evaluate and rank alternatives through normalization, weighted scoring, and compromise-based ranking. Six key criteria are considered: water availability, land suitability, flood risk, market access, government support, and microclimate conditions. The integration of PIPRECIA and MACROS enables a systematic and transparent evaluation process. The results indicate that location "G" achieves the highest compromise score of 0.985, signifying its suitability as the most optimal site for sustainable agricultural development. The primary contribution of this research lies in offering a quantitative and structured approach that accommodates environmental uncertainties while enhancing decision-making transparency. By integrating expert judgment with computational assessment, this model supports data-driven decision-making in the planning of agricultural development. These findings are expected to provide strategic insights for policymakers in formulating adaptive agricultural policies, strengthening food security, and improving farmer welfare through accurate and sustainable location selection.

Evaluasi Usability Aplikasi Presensi Mobile UMP Menggunakan Metode User Experience Questionnaire dan Think Aloud

2025-07-29T16:04:40+07:00

The implementation of a mobile-based digital attendance system in higher education aims to enhance the efficiency of academic administration and the accuracy of attendance data. However, various challenges remain that impact the quality of user experience, including suboptimal authentication processes, the absence of logout notifications, limited attendance data visualization, and lack of login information storage features. These issues can potentially diminish user convenience and reduce long-term application efficiency. At Universitas Muhammadiyah Purwokerto, a barcode-scanning-based attendance application has been deployed; however, a comprehensive usability and user perception evaluation has not yet been conducted. This study aims to evaluate the usability of the application using a mixed-method approach that combines quantitative and qualitative methods, namely the User Experience Questionnaire (UEQ) and the Think Aloud protocol. The UEQ method is employed to assess user perception across six dimensions: attractiveness, clarity, efficiency, dependability, stimulation, and novelty. In contrast, the Think Aloud method explores users’ real-time verbal feedback and difficulties while interacting with the application. A total of 100 respondents completed the UEQ, and 5 participants took part in the Think Aloud sessions. The results indicate that all UEQ dimensions received positive scores, with clarity scoring the highest (mean = 1.95) and novelty the lowest (mean = 1.35). Findings from the Think Aloud sessions corroborate the UEQ results and reveal previously unidentified areas for improvement. Based on these insights, user-centered design recommendations were developed and translated into a user interface prototype. This study not only provides a comprehensive usability evaluation but also offers direction for the development of a more intuitive and adaptive mobile attendance application that can be adopted by other educational institutions.

Question Answering System Zakat dengan Metode Long Short-Term Memory (LSTM)

2025-07-31T21:07:40+07:00

Zakat is a fundamental pillar of Islamic finance that serves as a mechanism for wealth redistribution. However, there is currently no Indonesian-language Question Answering System (QAS) capable of automatically and contextually responding to zakat-related queries. This study aims to develop a zakat-focused QAS using a Long Short-Term Memory (LSTM) model integrated into the Telegram platform. The dataset was compiled from the official BAZNAS zakat guidebook and processed through tokenization, padding, and label encoding. The model architecture consists of an embedding layer, two stacked LSTM layers (with return sequences, dropout, and recurrent dropout), followed by two dense layers (200 and 100 units) with additional dropout layers before the softmax output. The model was trained using the Adam optimizer (learning rate 0.003), a batch size of 24, and 100 epochs. Evaluation was conducted using a confusion matrix, resulting in a validation accuracy of 93%, with a precision of 0.94, recall of 0.93, and F1-score of 0.92 (weighted average). The system was deployed via the Telegram Bot API and demonstrated response times under two seconds, with stable performance across hundreds of question labels. This work contributes to the advancement of digital zakat education and presents a scalable solution that can be further extended within the ecosystem of Islamic Finance Technology and Digital Religious Education.

Implementasi Algoritma Apriori dalam Menemukan Pola Asosiasi pada Data Penjualan Produk Retail

2025-08-03T00:13:03+07:00

This study aims to implement the Apriori algorithm in finding association patterns in retail product sales data, using the Association Rule Mining approach. Evaluating the ruler or association rules formed based on the support, confidence, and lift parameters, in finding association patterns in retail product sales data with a focus on the relationship between product categories. The data used consists of 500 sales data as sample data and 5,972 transactions as test data. The data mining process was carried out on the main product categories such as Milk, Coffee, Tea, Drinks, Detergent, and Biscuit/Snacks, to find association rules that appear simultaneously with the Bulk Products category in one transaction time. The minimum support parameter was set at 0.02 and the minimum confidence was set at 0.5. By using these parameters, several significant association rules were obtained. One of the strongest rules shows that if products in the Milk, Coffee, Tea, Drinks, Detergent, and Biscuit/Snacks categories are purchased together, then there is a 64.3% probability (confidence) that products in the Bulk Products category are also purchased at the same time. The support value of this rule reached 3.8%, and the lift value was 1.49, indicating a positive association and not a coincidence. Evaluation of the test data showed that this pattern was consistently found across 5,972 transactions, with a repeatability rate of 61.7%. The results of this study demonstrate that the Apriori algorithm is effective in identifying consumer purchasing patterns that can be utilized for product placement strategies, bundling offers, and inventory planning in retail management.

Perancangan Sistem Informasi Pengelolaan Akun Pada Aplikasi SAP dengan Metode Waterfall

2025-06-28T10:52:32+07:00

The need for an account management application in the SAP system is crucial for maintaining the reliability of annual audit trails, which are an essential part of corporate data governance. This application is designed to facilitate access rights management and record every user activity within the SAP system, aiming to ensure the integrity and security of company data. Without a robust system for managing access rights and monitoring user activity, audit trails become vulnerable to risks such as data misuse, shifting of responsibilities, and invalid user credentials. This can lead to financial and reputational losses for the company. This research aims to design and build an effective SAP system account management application to enhance audit trail reliability and minimize risks that can occur in the SAP system's account management process. A good information system will enable companies to manage and secure data more effectively, support license audits, and meet applicable compliance standards. The Waterfall method is used to achieve these objectives, encompassing requirements analysis, system design, implementation, testing, and maintenance phases. Through this approach, the expected outcome is an SAP account management application that is effective, secure, and supports audit standards as well as corporate data security governance.

Analisis Pengaruh Technology Readiness Terhadap Minat QRIS Dalam Melakukan Pembayaran Dengan Menggunakan Metode TRAM

2025-07-25T22:41:47+07:00

QRIS is a payment method that is currently frequently used by the public in making transactions. QRIS is currently quite popular among the public, as evidenced by the increasing number of QRIS users. The use of QRIS is highly sought after by the public. This interest in using QRIS is believed to occur due to the influence of the application of new technology (Technology Readiness Acceptance Model). The purpose of this study was to determine the factors that influence interest in using QRIS and to determine the obstacles faced by the public in making payments using QRIS based on the Technology Readiness Acceptance Model method. The analysis method used is the Partial Least Squares-Structural Equation Modeling method. The results of the study indicate that the factors that influence interest in using QRIS in making payments based on the Technology Readiness Acceptance Model method are optimism, innovativeness, perceived ease of use, and perceived usefulness. Meanwhile, obstacles faced in making QRIS payments are a lack of security, frequent failures in making transactions, and not all business actors have QRIS in payment transactions.

Evaluasi Kepuasan Pelanggan Dalam Berbelanja Online Pada Aplikasi Shopee Menggunakan Metode Importance Performance Analysis

2025-08-02T01:17:24+07:00

Shopee is one of the most popular e-commerce platforms in Indonesia, especially among university students. At Malikussaleh University, it has become the preferred application for meeting both academic and personal needs. However, despite its high usage, there are still complaints regarding unsatisfactory service quality, such as delayed deliveries, difficulty in processing returns, and slow customer support responses. This study aims to analyze customer satisfaction with Shopee services using the Importance Performance Analysis (IPA) method to assess the alignment between service importance and performance. The study evaluates four key dimensions: Web Design, Fulfillment, Customer Service, and Security/Privacy. The results show that the average conformity level (TKI) across 17 service attributes is 94.59%. A TKI value below 100% suggests that the provided services have not fully met customer expectations. Moreover, the IPA quadrant analysis reveals that several attributes—such as delivery punctuality, responsiveness to complaints, and clarity of product information—fall into Quadrant A. Attributes in this quadrant are critical and underperforming, indicating top priorities for immediate improvement to enhance overall service quality.

Analisis Klasifikasi Kesiapan Digital Desa Menggunakan Decision Tree dan Pemetaan Spasial

2025-08-08T23:03:12+07:00

Digital transformation at the village level is a strategic element in promoting equitable development and improving public service delivery. However, the level of digital readiness across regions remains uneven. This study aims to classify the digital readiness of villages in West Java Province by utilizing data from Open Data Jabar (opendata.jabarprov.go.id) related to the number of digital villages, internet access, and village development strata. A Decision Tree classification algorithm was employed to categorize regions into two readiness classes: high and low. The modeling results indicate that the number of self-reliant (mandiri) villages and the percentage of villages with internet access are the most influential variables in the classification. Although internet infrastructure is available in most areas, it does not always correspond to the level of village digitalization. Districts with high internet access but a low number of self-reliant villages are still classified as having low readiness. The model achieved an accuracy of 83%, although its performance in identifying the high readiness class was limited due to class imbalance in the dataset. Spatial visualization was also used to highlight regional disparities in digital readiness. This study provides an early contribution to digital readiness mapping of villages using a machine learning approach in Indonesia.

Aplikasi Three Tier Sistem Informasi Manajemen Kepegawaian Menggunakan Model Prototype

2025-07-31T20:59:58+07:00

Human resources is a service department that assists employees and leaders of an organization, as well as a division that handles individual or personal issues that, when applied to an organization, concern employees or staff. Balai Pengkajian dan Pengembangan Komunikasi dan Informatika (BPPKI) Surabaya has personnel data that needs to be stored and processed in order to facilitate computerized storage and reporting. The implementation of a web service information system is a collection of functions or methods contained on a server that can be called by clients remotely. Clients can be users who use the desktop application implemented in this study, or the web as a mobile communication device. The implementation of the Three Tier application model architecture is in terms of infrastructure and documents used as data exchange formats. The Human Resources Management Information System application was developed using the Three-Tier desktop-based method with the prototype method. The Human Resources Management Information System application consists of four key components: the transaction process information system for periodic salary increases, salary, rank promotions, and leave requests; simplifying report generation; and minimizing data input errors. The implementation and results of black box testing show that all form features tested on the human resources management system functioned successfully and as expected, making this application suitable for implementation to assist human resources management at Balai Pengkajian dan Pengembangan Komunikasi dan Informatika (BPPKI) Surabaya.

Klasifikasi Tingkat Kepuasan Peserta Pelatihan Balai Besar Pelatihan Vokasi dan Produktivitas Menggunakan Algoritma C5.0

2025-08-06T21:38:37+07:00

The evaluation of training participant satisfaction at the Center for Vocational Training and Productivity Development (BBPVP) has traditionally relied on conventional methods, resulting in less accurate and unstructured outcomes. The core issue necessitates a data-driven solution to enhance objectivity and reliability. This study aims to develop a C5.0 algorithm-based classification model to automatically measure participant satisfaction levels and identify dominant influencing factors. The methodology includes collecting survey data from 300 respondents across five SERVQUAL attributes (reliability, assurance, responsiveness, empathy, tangibles), data preprocessing, dataset splitting (80:20), and model development using Python’s Scikit-learn library. Results indicate a model accuracy of 98.3% (12% higher than Naïve Bayes), with "assurance" as the most influential attribute (gain ratio: 0.638). Contributions of this research include: (1) providing BBPVP with an accurate data-driven satisfaction evaluation tool, (2) offering strategic recommendations to improve training quality, particularly in assurance, and (3) potential adoption of this method as a national vocational training evaluation standard.

User Interface Design for Doctor Reservation Website using Design Thinking Method

2025-06-06T22:45:24+07:00

In the current era of digital transformation, the rapid integration of healthcare services is crucial to meet patient expectations and improve service delivery. Klinik Putri faces challenges such as long queues and difficulties in booking doctor appointments, which negatively impact patient satisfaction. This study aims to design a user-friendly, web-based doctor reservation system using the Design Thinking methodology, which consists of five stages: empathize, define, ideate, prototype, and test. Usability testing was conducted using the System Usability Scale (SUS), a standardized tool for evaluating system usability. The results showed an average SUS score of 83, placing the system in the “acceptable” category, Grade B, and receiving an “excellent” rating according to the Adjective Rating scale. These findings demonstrate that the proposed website design effectively addresses user needs, enhances the user experience, and contributes to improving the efficiency of healthcare services at Klinik Putri.

Optimalisasi Akurasi Prediksi Curah Hujan Bulanan Menggunakan Deep Learning

2025-08-14T12:27:55+07:00

The Province of Lampung exhibits high rainfall variability influenced by various atmospheric dynamics such as the Asian Monsoon, Australian Monsoon, El Niño–Southern Oscillation (ENSO), and the Indian Ocean Dipole (IOD). Accurate rainfall prediction is crucial across multiple sectors, including agriculture, water resource management, and hydrometeorological disaster mitigation. However, prediction methods commonly used in the region are still dominated by statistical approaches or conventional machine learning techniques, which often struggle to capture long-term temporal patterns in rainfall data. On the other hand, deep learning technologies such as the Recurrent Neural Network (RNN) and Gated Recurrent Unit (GRU) offer better capabilities in modeling time series data, yet no specific comparative evaluation has been conducted for rainfall prediction in the Lampung Province. Comparing these two methods is important because the architectural characteristics of RNN and GRU differ in handling long-term dependencies, and selecting the right model can directly impact prediction accuracy and the effectiveness of decision-making in affected sectors. This study aims to implement and compare the performance of RNN and GRU in predicting monthly rainfall in Lampung Province using data from 80 rain gauges distributed across 15 districts/cities over the period from January 1991 to February 2025. The results show that the RNN model outperforms the GRU model, with lower RMSE (115.61 vs. 119.50), smaller MAE (86.94 vs. 91.28), and higher R² (0.35 vs. 0.30). Predictions for the period from March 2025 to February 2026 reveal a clear seasonal pattern, with minimum rainfall occurring in August 2025 (peak dry season) and maximum rainfall in January 2026 (peak rainy season). This study demonstrates that RNN is more effective than GRU in capturing the temporal patterns of rainfall, making it more recommended for long-term prediction applications.

Penerapan Algoritma BM25 dalam Pencarian Lowongan Pekerjaan pada Website Job Portal

2025-08-16T21:30:26+07:00

The development of the digital era has grown rapidly all the time which has significantly changed the job search process for job applicants, making Online Job Portals one of the main places in human resource recruitment activities, however, the effectiveness of Job Portals Job search still has fundamental weaknesses such as the job search technology used still uses simple string matching which can cause less relevant search results and reduce the quality of user experience in applying for jobs. This study was conducted to improve the quality of job vacancy search results on Job Portal A Career by applying the Okapi BM25 algorithm. This research method uses a Rapid Application Development (RAD) development approach, such as designing a client server architecture with Next.js as the frontend, ASP.NET Core as the backend and PostgreSQL as the main database. The BM25 algorithm is integrated directly into the database using the VectorChord BM25 extension to calculate the search relevance score with the user inputted query. In testing with the query “accelist the quality support career IT need”, the system displays 800 of 1,011 documents (79.13%) with a non-zero relevance score. Furthermore, evaluation through User Acceptance Testing (UAT) showed a user satisfaction rate of 91.2%, confirming that BM25 is capable of displaying the most relevant results at the top of the rankings and supporting the effectiveness of the search system. The results of this study can be concluded that the BM25 algorithm is a more effective and efficient search solution with high scalability potential for application to other web-based job search systems.

Implementasi Animasi 2D menggunakan Motion Graphic sebagai Media Informasi Palang Merah Indonesia

2025-08-13T18:26:04+07:00

The Indonesian Red Cross (PMI) has a significant responsibility in conveying humanitarian information to the public. However, the challenges of effectively communicating information to the broader community remain a primary concern, as conventional media often lack appeal and are difficult to comprehend fully. To address this issue, this study implements 2D animation based on motion graphics as a communication medium for PMI. The animation production process includes creating storyboards, developing visual illustrations, graphic processing, adding motion elements, voice narration, and audio-visual synchronization, resulting in a communicative and engaging medium. 2D animation was chosen because it can present messages with simple yet clear visuals, while motion graphics provide engaging motion dynamics that make information easier to understand and remember. The integration of both allows for the delivery of messages that are concise, interactive, and in line with the characteristics of digital media that are widely accessed by the public. Evaluation results show a significant increase in the level of understanding among respondents after watching the video, with post-test scores reaching 94.8% in the PMI member group and 91.2% in the general public group. These findings affirm that 2D animation media based on motion graphics is effective in enhancing the appeal, understanding, and effectiveness of PMI communication, thus it can be an innovative alternative strategy to expand the reach of humanitarian information.

Optimisasi VGG16 dengan Transfer Learning dalam Mendeteksi Penyakit Pada Daun Jagung

2025-06-18T00:10:19+07:00

Corn is one of the major agricultural commodities that plays a strategic role in national food security. However, its productivity often declines due to leaf diseases such as Blight, Common Rust, and Gray Leaf Spot. Manual disease detection is considered inefficient and prone to human error, especially on a large scale. This study aims to develop an automated deep learning-based system for accurate classification of corn leaf diseases. The proposed model utilizes the Convolutional Neural Network (CNN) architecture VGG16 with a transfer learning approach. The dataset comprises 1,200 labeled images of corn leaves categorized into four disease classes, obtained from Kaggle. Image augmentation techniques were applied to improve data diversity and enhance model generalization. The performance of VGG16 was compared with VGG16 Baseline architecture and MobileNetV2. Experimental results show that VGG16 with transfer learning achieved the highest classification accuracy of 96.25%, outperforming the baseline VGG16 (92.92%) and MobileNetV2 (84.58%). These findings demonstrate the effectiveness of VGG16-based transfer learning in automating corn leaf disease detection, supporting the implementation of precision agriculture technology.

Klasifikasi Tingkat Risiko Gempa di Indonesia Menggunakan Pola Spasial dan Temporal Berbasis Decision Tree

2025-06-17T20:52:54+07:00

Indonesia is an area that is very vulnerable to earthquakes due to its location in the meeting zone of active tectonic plates. This study aims to classify the level of earthquake risk based on spatial and temporal patterns using the Decision Tree method as a solution in predicting potential earthquake hazards. The data used is earthquake data in Indonesia from 2015 to 2023 obtained from public datasets, including location information (latitude and longitude), event time (year and month), and earthquake magnitude. Earthquakes are categorized into three risk classes: Low (M < 4.0), Medium (4.0 ? M < 6.0), and High (M ? 6.0). The Decision Tree model was successfully built with an average accuracy of 88% on the test data. The results show that earthquakes mostly occur in active subduction zones such as the Sunda Subduction Zone (Sumatra and Java), Banda Arc (Nusa Tenggara, Maluku, Seram), Sulawesi, and Papua. Temporal analysis also shows fluctuations in the number of earthquakes by year and season, with increased activity in certain months. The spatial visualization reinforces the finding that the eastern region of Indonesia is more seismically active than the western region. This research proves that machine learning approaches can be used to support earthquake disaster mitigation through historical data-based risk identification.

Analisis Sentimen Publik terhadap ‘Save Raja Ampat’ di Media Sosial Menggunakan Model IndoBERT

2025-06-15T22:36:07+07:00

The "Save Raja Ampat" campaign has emerged as a significant environmental issue that has garnered widespread public attention on social media platforms, particularly TikTok and YouTube. Videos tagged with #SaveRajaAmpat have sparked various public responses, ranging from full support to criticism of natural resource exploitation. This phenomenon highlights the importance of understanding public sentiment as an indicator of the campaign's effectiveness. This study aims to analyze public sentiment toward the campaign using a language modeling approach based on artificial intelligence, namely IndoBERT. The data were obtained from user comments on TikTok videos promoting the “Save Raja Ampat” campaign, totaling 10,000 comments. The analysis process involved several stages, including data preprocessing, sentiment labeling (positive, negative, neutral), and the training and evaluation of the IndoBERT model. Preliminary results indicate that the majority of public sentiment toward the campaign is positive, with the model achieving an accuracy rate of 71% in sentiment classification. This study contributes to understanding public perception of environmental issues and demonstrates the effectiveness of using the IndoBERT model in the context of social media.

Implementasi Metode Analitycal Hierarchy Process dan Multi-Objective Optimization by Ratio Analysis Untuk Rekomendasi Laptop

2025-08-15T22:51:20+07:00

Laptops have become essential in the world of work, education, and society. With a laptop, tasks such as creating reports, sending data, learning, and even entertainment become easier. However, the variety of laptops available with different specifications can confuse people when choosing one that suits their profession and status. This confusion often leads to wasted time and the risk of choosing a laptop that does not meet their needs. Therefore, a decision support system (DSS) is needed to provide laptop recommendations based on desired criteria. In this study, the method used is a collaboration of Analytical Hierarchy Process (AHP) and Multi-Objective Optimization By Ratio Analysis (MOORA). AHP is used to calculate the weight of laptop criteria according to desired criteria, while MOORA is used to rank the recommended laptop values suitable for use. The implementation of the AHP and MOORA methods in this study resulted in laptop recommendations that meet the desired criteria and specifications of the community. Based on manual calculations in this study, the top-ranked laptop recommendation is alternative A8, the HP Victus Gaming Laptop 15 with a Yi of 0.424, followed by alternative A2, the HP Pavilion Gaming 15 with a Yi of 0.382. This study is considered successful because the results of manual calculations and those of the system built are consistent. Thus, the implementation of AHP and MOORA methods in a web-based system can be used for laptop recommendations.

Prediksi Kegagalan Perangkat Industri Menggunakan Random Forest dan SMOTE untuk Pemeliharaan Preventif

2025-08-19T21:06:43+07:00

Preventive maintenance is an essential strategy to minimize losses due to industrial equipment failures. This study aims to develop an equipment failure prediction model using the Random Forest algorithm with the SMOTE technique to address class imbalance. The dataset used is the AI4I 2020 Predictive Maintenance Dataset with 10,000 entries and six main input variables. Preprocessing includes normalization of numerical features, one-hot encoding for categorical features, and handling of missing values. The Random Forest model was optimized using GridSearchCV and compared with K-Nearest Neighbors. Results show that Random Forest with SMOTE achieved 97% accuracy, 0.47 precision, 0.75 recall, and 0.58 F1-score on the failure class. This model outperforms KNN in detecting failures, particularly in imbalanced data. These findings contribute to the development of an early warning system to support preventive maintenance in industrial environments.

Model Deep Learning Berbasis Multilayer Perceptron untuk Identifikasi Demam Berdarah Dengue dan Tifus

2025-08-21T19:59:24+07:00

Dengue Hemorrhagic Fever (DHF) and Typhus/Typhoid are two infectious diseases often found in tropical areas. In Indonesia, data shows that cases of DHF and typhoid are quite high, so a system is needed that can help doctors make faster and more accurate decisions based on blood test results. Based on the previous explanation, this study aims to apply the Deep Learning Multilayer Perceptron (MLP) method to be able to identify dengue fever and typhus. This study uses a Deep Learning-based Multilayer Perceptron approach for accurate classification of Dengue Fever, Typhoid Fever, and Normal cases using clinical blood parameters and selected symptoms. This methodology consists of several stages: dataset acquisition, preprocessing, model architecture design, training, and evaluation. The dataset was taken from Dumai City Hospital medical record data from 2023 to 2024, totaling 379 patient data used to identify Dengue Fever and Typhus using 7 clinical parameters as the main input obtained from laboratory examination results and patient clinical symptoms: Hemoglobin, Leukocyte, Platelet count, Hematocrit level, Headache, Abdominal pain, and diarrhea. Based on the results obtained, the application showed the best performance in classifying Dengue Fever, which is shown through the achievement of the model evaluation metrics as follows. The test results indicate that an increase in the amount of test data is directly proportional to the percentage of classification success achieved by the system. Based on the test results with 10% validation data, 70 % training data, and 20 % test data, the system showed very good performance with an overall accuracy of: 98.68% (Accuracy = 0.9868), which indicates a high level of success in classifying for the three classes, namely Normal, Dengue Fever, and Typhus.

Analisis Kepuasan Masyarakat Terhadap Proses Pengurusan Sertipikat Analog Ke Elektronik Menggunakan Metode Naïve Bayes

2025-08-21T20:51:22+07:00

The certificate media conversion program from analog to electronic implemented by the Ministry of ATR/BPN in Sejati Village requires evaluation to ensure its effectiveness. The main problem faced is the limited use of quantitative, data-driven analysis in identifying the factors that influence public satisfaction. This study aims to analyze the level of public satisfaction using the Naïve Bayes method to classify and predict the influence of related variables. Data were obtained from 250 respondents through questionnaires based on digital public service indicators, covering demographic variables, perceived benefits, obstacles, support, service speed, and procedural simplicity. The results show that the level of public satisfaction is in the high category, with procedural simplicity and service speed proven to be the most significant variables influencing satisfaction prediction. The Naïve Bayes model achieved an accuracy of 94%, demonstrating its effectiveness in predicting satisfaction levels. These findings serve as a basis for improving policies and strategies to enhance the quality of digital public services, particularly in the implementation of electronic certificate media conversion in the future.

Analisis Algoritma K-Means Clustering Dalam Pengelompokan Prestasi Belajar Siswa Menengah Atas (SMA)

2025-08-21T21:10:16+07:00

The increased use of social media among high school students has a positive and negative impact on academic achievement. This can be seen from changes in learning patterns, concentration levels, and students' motivation in participating in learning activities. This study aims to classify student learning achievement based on the level of social media use using the K-Means Clustering algorithm. K-Means Clustering is one of the main methods in data mining. which is a technique of grouping data based on the similarity of its characteristics. The parameters used in analyzing this study are Social Media Duration (X1), Active Time (X2), Main Platform (X3), Main Goal (X4), Social Media Access Time While Learning (X5), Social Media Addiction (X6), Social Media Addiction Level (X7), Number of Study Groups (X8) and Academic Average (X9). Based on the K-Means Clustering method, it has been proven to be able to group students based on the level of social media use. These results can be seen from the cluster category C0 (High) with 46 students, C1 (medium) with 80 students, and C2 (Low) with 72 students. The contribution of this research benefits students by helping them understand the relationship between social media usage habits and learning achievement, so as to encourage more effective time management.

Analisis Klaster Penyebaran Berat Produk Mesin Sachet Menggunakan Metode Algoritma K-Means

2025-08-21T22:28:58+07:00

This study aims to analyze the distribution patterns of sachet machine product weights using the K-Means algorithm as a clustering technique. The dataset consists of 940 entries of primary production records, each containing ten weight measurement samples per production cycle. The data underwent a cleaning process to ensure the absence of missing values, duplicates, and outliers, followed by the selection of relevant attributes (product weight samples) and transformation using Min-Max normalization to scale all variables within the 0–1 range. The clustering process was performed iteratively by updating the centroids until convergence was achieved. The evaluation results indicate that the optimal number of clusters is three (k=3) with a Silhouette Coefficient of 0.55, reflecting a good balance between intra-cluster homogeneity and inter-cluster separation. Cluster 1 represents products with relatively low weights (8.00–8.18 grams), Cluster 2 includes medium-weight products (8.19–8.34 grams), and Cluster 3 consists of high-weight products (8.36–8.98 grams). Overall, the product weights tend to be stable with low variation, although some anomalies were observed in certain machines. These findings demonstrate that the K-Means algorithm can effectively classify product weight data, providing valuable insights for quality control, product variation identification, and minimizing risks of deviation from production standards.

Optimasi Algoritma Random Forest untuk Prediksi Eksport Kelapa Sawit Global

2025-08-21T22:33:51+07:00

Palm oil production is a strategic commodity in global trade, with a trend showing an increase from year to year. This study aims to optimize the Random Forest algorithm in predicting the amount of global palm oil production based on historical data. The dataset used consists of 12,458 observations with one dependent variable (Palm_Oil_00002577_) representing the amount of palm oil production, and four independent variables: country, Code, Year, and Palm_Oil_00002577_log. The data is divided into 80% for training (9,966 observations) and 20% for testing (2,492 observations). The model optimization process is carried out by adjusting the key parameters of Random Forest using Grid Search and Cross-Validation. The initial Random Forest model (without optimization) produces a Root Mean Squared Error (RMSE) value of 115.27 and an R-squared (R²) value of 0.9824 on the test data. After optimization using Grid Search and Cross-Validation on key parameters (n_estimators, max_depth, and max_features), the optimized model showed significant performance improvements, with the RMSE decreasing to 103.54 and the R² increasing to 0.9984. The decrease in the RMSE indicates a reduction in the model's average prediction error, while the increase in R² approaching 1 indicates the model's ability to explain almost all of the variation in global palm oil production data. These results indicate that parameter optimization in Random Forest can substantially improve prediction accuracy, enabling the model to be used as a production planning tool and strategic decision-making tool in the palm oil commodity trading sector.

Prediksi Jumlah Kebutuhan Biji Kopi Berdasarkan Pola Konsumsi Konsumen dengan Algoritma Apriori

2025-08-21T21:04:22+07:00

Coffee bean prediction is needed for optimal inventory management to maintain efficiency. This data grouping is taken from customer shopping consumption patterns. Based on the research aims to predict the amount of coffee bean needs based on consumer consumption patterns by applying the Apriori algorithm. Utilization of processed transaction data can provide what steps should be taken in the future. Based on this, this study aims to predict the amount of coffee bean needs based on consumer consumption patterns with the Apriori algorithm. The Apriori algorithm forms association rules based on a combination of data indicators used. These data indicators are sourced from Freehand Coffee. Based on the use of the Apriori algorithm in predicting coffee bean needs based on consumer consumption patterns, the results showed that the Apriori algorithm is able to provide product recommendations in the form of associative or consumer transaction patterns by collecting transaction data and then experimenting with existing data indicators. The contribution of this research can help Freehand Coffee to estimate coffee bean needs and optimize stock management, this research also helps in selecting drinks based on consumer consumption.

Perbandingan Metode MAUT dan Profile Matching Terhadap Sistem Pendukung Keputusan Seleksi Calon Paskibraka

2025-08-21T20:00:48+07:00

The selection of candidates for the Flag Raising Troop (Paskibraka) requires an objective, transparent, and consistent assessment system based on physical criteria, general intelligence, personality, marching skills, and physical fitness. This study compares the Multi Attribute Utility Theory (MAUT) and Profile Matching (PM) methods in supporting selection decision making. The MAUT method produces a utility score by considering the weight of each criterion, while PM assesses the level of conformity to the ideal profile through GAP analysis grouped into Core Factor (CF) and Secondary Factor (SF). The purpose of this study is to compare the accuracy of the MAUT and Profile Matching methods in the decision support system for Paskibraka candidate selection. The calculation results show that candidate A3 obtained the highest score in both methods (MAUT = 0.970; PM = 3.600), followed by A1 (MAUT = 0.957; PM = 3.500) and A9 (MAUT = 0.918; PM = 3.450). This consistency indicates a convergence in the assessment of high-performing candidates. However, differences emerge in the middle ranks, for example, A2, which ranked 8th on the MAUT (0.856) but rose to 4th on the PM (3.100). Differences in assessment principles are the main factor: MAUT emphasizes an even distribution of scores across all criteria, while PM focuses more on fit with the ideal profile, particularly on the core criteria (CF). This finding emphasizes that the choice of methods should be tailored to selection priorities, or used in combination to obtain more accurate and objective results.

Analisis Algoritma K-Means Clustering untuk Pengelompokan Rekomendasi Judul Proposal Tugas Akhir Mahasiswa

2025-08-21T21:04:33+07:00

The academic process requires speed and accuracy in processing student data, such as submitting final project titles. In the context of final project title recommendations, many universities have not yet implemented the Data Mining approach optimally. Based on this, this study aims to recommend grouping of student final project proposal titles. The K-Means clustering method can be used in grouping data based on similarities between analyzed objects. With the K-Means method, the student grouping process utilizes grade data from the courses of Rock Mechanics, Drilling and Excavation Techniques, Underground Mining Methods, Reserve Modeling and Evaluation, Explosives and Blasting Techniques, Open Pit Mining, Mine Drainage Systems, Mapping Surveys, and Mineral Resources. The results of K-Means are strongly influenced by the k parameter and centroid initialization. The research variables include data mapping of course grades of students in the Mining Engineering Study Program. Based on the K-Means Clustering Method, it has been able to divide 104 student value data into 3 clusters, namely Natural Resource Exploration (C0), Geomechanics (C1) and Mining Environment (C2). The results of Cluster CO are 60, the results of Cluster C1 are 27 and the results of Cluster C2 are 17. The contribution of this research can provide fast, precise and accurate information in grouping recommendations for student final project proposal titles.

Deteksi Pelanggaran Tata Tertib Siswa Sistem Cerdas Menggunakan Face Recognition dengan Metode Convolutional Neural Network

2025-08-21T21:08:57+07:00

Student disciplinary violations are a social problem increasingly common in schools and can negatively impact students' academic and moral development. This phenomenon requires an effective identification system so that prevention and mitigation efforts can be carried out quickly and accurately. This research aims to develop a student face detection system based on Digital Image Processing (DIP) technology that functions to identify and classify adolescent disciplinary violations. The designed system utilizes a camera as an image acquisition device, then processes it to detect the presence of student faces in real-time. The face detection process is carried out using the Haar Cascade Viola-Jones method, which is known to be able to recognize faces with high speed and accuracy. Once a face is detected, the system continues the analysis process using the Convolutional Neural Network (CNN) method to classify facial expressions and behavioral patterns that could potentially indicate violations. The integration between Haar Cascade and CNN allows the system to work efficiently in identifying signs of negative behavior based on visual data. System testing shows satisfactory results, with a high level of facial detection accuracy and fairly reliable behavior classification capabilities. This technology has the potential to be used as a monitoring tool in the school environment, allowing teachers and school management to quickly identify students who need special attention. With the implementation of this system, it is hoped that schools will be able to provide timely guidance, prevent the escalation of deviant behavior, and create a more conducive learning environment. The use of digital image processing-based technology for detecting and classifying student behavior is a relevant innovation in the modern education era, while also supporting efforts to prevent juvenile disciplinary violations through a systematic and measurable approach.

Analisis Data Mining dengan Metode K-Means Clustering Dalam Pengelompokan Penggunaan Alat Kontrasepsi

2025-08-21T20:56:58+07:00

Family Planning (KB) is a strategic government effort to suppress population growth and improve the quality of life. The availability of various types of contraceptives can delay unwanted pregnancies, including in women facing increased pregnancy risks. Based on this, this study aims to cluster contraceptive use. The K-Means Clustering method is an unsupervised learning algorithm used to group data into several clusters based on similar characteristics. This algorithm works by minimizing the distance between the data and the cluster center (centroid). The advantages of K-Means are its simplicity and speed in processing large data. This research variable uses data from the 2024 Family Data Collection of the BKKBN Representative Office of West Sumatra Province in West Pasaman Regency. Based on the application of the K-Means Clustering method to the contraceptive use data, the grouping is obtained into three clusters: low use of MKJP contraceptives, moderate use of MKJP contraceptives, and high use of MKJP contraceptives. This study contributes in the form of a data mining-based analysis model that is able to group contraceptive use patterns in a more structured and objective manner. By applying the K-Means Clustering method, this study produces information that can be used to identify the characteristics of each user group, so that relevant agencies can design more targeted contraceptive counseling and distribution strategies.

Analisis Cluster Algoritma K-Means Untuk Pengelompokan Kondisi Gizi Balita Pada Posyandu

2025-08-26T18:29:30+07:00

Toddler health is a crucial indicator of community and national development. Integrated Service Posts (Posyandu) play a key role in monitoring the nutritional status of toddlers through routine weight and height checks. This study aims to analyze toddler nutritional status using the K-Means Clustering algorithm, a non-hierarchical method that groups data based on centroid proximity. The data came from 98 toddlers at the Posyandu in Manggung Village, North Pariaman District, Pariaman City, including weight, height, weight-for-age, height-for-age, weight-for-height, and weight gain. The K-Means results showed a distribution of three clusters: C0 (undernourished) with 37 toddlers, C1 (severely malnourished) with 17 toddlers, and C2 (well-nourished) with 44 toddlers. The majority of toddlers were categorized as well-nourished. This research contributes to the rapid identification of toddler nutritional problems, enabling Posyandu staff to take appropriate preventive and corrective measures.

Identifikasi Varietas Kopi Berdasarkan Analisis Warna dan Tekstur Menggunakan Metode Convolutional Neural Network

2025-08-21T20:41:26+07:00

Coffee is a plantation commodity with high economic value in Indonesia, with various varieties such as Arabica, Robusta, and Liberica. Differences in coffee varieties can generally be identified through the physical characteristics of the beans, especially color and texture. Based on this, this study aims to develop a digital image-based coffee variety identification system using the Convolutional Neural Network (CNN) method with color and texture analysis as the main features. The research stages include coffee bean image acquisition, pre-processing including color segmentation and image conversion to grayscale, and color and texture feature extraction. This research dataset comes from images of unroasted coffee beans, commonly called green beans, taken using a high-resolution smartphone camera and also using secondary data taken from the Kaggle site. Both types of datasets have the same characteristics and resolution to maintain data consistency. The image dataset is divided into training data and test data, then used to train and test the Convolutional Neural Network (CNN) model. Based on this study, the Convolutional Neural Network (CNN) method can identify coffee varieties based on color and texture analysis. By using 210 training data and 90 test data of coffee bean images, the CNN method can produce an accuracy rate of 94,44%. This research contribution has the potential to be a supporting solution in the process of identifying coffee varieties quickly, accurately, and consistently, so that it can help the coffee industry in the sorting and quality control process.

Analisis Metode Forward Chaining dan Certainty Factor untuk Diagnosa Penyakit pada Ibu Hamil

2025-08-21T20:52:40+07:00

The high number of complications that occur during pregnancy and childbirth has the potential to significantly increase the risk of morbidity and mortality in pregnant women. The Maternal Mortality Rate (MMR) reflects the condition of pregnant, delivering, and postpartum mothers, which remains relatively high and is a major concern in the health sector. Based on this, this study aims to develop and evaluate an Expert System based on the Forward Chaining and Certainty Factor methods to diagnose diseases in pregnant women at an early stage, thereby providing fast and accurate medical decision support and minimizing the risk of complications during pregnancy. The Forward Chaining and Certainty Factor methods were chosen for their ability to handle rule-based inference processes and provide certainty level calculations in the diagnosis results. Forward Chaining is used to find solutions based on the symptoms entered by users, while the Certainty Factor helps assign confidence weights to the generated diagnosis. The dataset in this study consists of 30 data samples with 30 types of symptoms experienced by patients as variables. The results show that the Forward Chaining and Certainty Factor methods are capable of producing disease diagnoses in pregnant women with an accuracy rate of 95%. The contribution of this research is to improve the quality of maternal health services through fast and accurate diagnoses by medical personnel and to assist pregnant women in obtaining an initial diagnosis of common diseases during pregnancy.

Business Intelligence untuk Validasi Desain Karakter Berbasis Budaya pada Game Aventala: “The Lost Tribe”

2025-05-25T23:16:10+07:00

This study focuses on culture-driven character design for Aventala: The Lost Tribe by transforming Indonesian endemic animals and cultural elements into humanoid forms documented in a character sheet. The objective is to formulate and validate a culture-driven character design pipeline via: (i) a personification sheet that maps physiology–fantasy–psychology–sociology, (ii) scene-based moodboards to align tone and persona, and (iii) a user study employing a six-indicator 5-point Likert instrument (mythology, culture, fantasy, naming, traits, and sheet readability) analyzed in a Business Intelligence dashboard. The method combines narrative comprehension (DRTA), qualitative data curation, sheet construction, moodboard development, and an online survey with the target audience. Results show a moodboard satisfaction level of 85.24% and character acceptance ranging from 83% to 86%, indicating coherence across cultural representation, fantasy elements, naming, and traits. These findings suggest the proposed pipeline is effective for evidence-based design, and the personification sheet serves as a practical cross-team artifact to guide iteration decisions.

Implementasi Simple Queue pada Router Mikrotik RB941-2nd untuk meningkatkan Quality of Service Jaringan Internet Kampus

2025-08-17T00:12:39+07:00

The internet has become a primary necessity in supporting various activities in higher education, including academic, research, and administrative purposes. The availability of stable and high-quality internet services is crucial, as most campus activities have shifted to digital platforms such as online learning, access to scientific journals, email communication, and cloud-based applications. Institut Teknologi dan Bisnis Indonesia (ITB Indonesia), located in Medan, provides internet infrastructure that serves hundreds of users daily, consisting of lecturers, students, and administrative staff. The high number of users creates network load, which may decrease service quality if not properly managed. Therefore, an appropriate bandwidth management method is required to maintain network stability and performance. This study implements the Simple Queue method on MikroTik devices to analyze internet traffic patterns and evaluate Quality of Service (QoS) parameters, including throughput, delay, jitter, and packet loss. The Simple Queue method is chosen for its ability to distribute bandwidth proportionally among users, preventing excessive usage by certain individuals. The results show an improvement in throughput, where lecturers increased from 855 Kbps to 1000 Kbps, staff from 855 Kbps to 1000 Kbps, and students from 854 Kbps to 941 Kbps. Delay values decreased from 103 ms to 80.7 ms for lecturers, from 103 ms to 67.7 ms for staff, and from 419 ms to 68.3 ms for students, indicating reduced transmission latency. Jitter values decreased significantly from 119 ms to 18.9 ms for lecturers, from 119 ms to 11.2 ms for staff, and from 296 ms to 21.8 ms for students, reflecting more stable data transmission. Packet loss also decreased from 14.5% to 0% for lecturers and staff, and from 14.6% to 5.90% for students, showing improved data delivery reliability. The findings indicate that the implementation of the Simple Queue method effectively improves internet performance in the campus network. Furthermore, the results can provide useful recommendations for network administrators in optimizing bandwidth management, developing network infrastructure, and formulating strategies to enhance internet service quality in higher education environments.

Penerapan Metode Fuzzy Sugeno Untuk Optimalisasi Persediaan Pakaian

2025-08-27T21:41:02+07:00

In the fashion retail industry, inventory management is a major challenge due to fluctuating and unpredictable customer demand. Errors in inventory planning may lead to overstocking or stockouts, increased storage costs, and decreased customer satisfaction. This study aims to develop a decision support system using the Sugeno fuzzy logic method to optimize clothing inventory. The input variables consist of initial stock, incoming goods, and outgoing goods, which are processed through fuzzification, inference, and defuzzification stages to produce the predicted final stock. Experimental results show that the Sugeno fuzzy model achieves better accuracy compared to conventional methods, with a Mean Absolute Percentage Error (MAPE) of 17.99%, equivalent to a prediction accuracy of 82.01%. The main contribution of this research lies in the application of the Sugeno fuzzy method to local fashion retail inventory management, which has generally been carried out manually. This approach enables the system to provide more precise stock recommendations, thereby helping stores reduce the risk of overstocking and stockouts, improve operational efficiency, and enhance business competitiveness.

Analisis Perbandingan Kinerja Algoritma K-Means dan K-Medoids dengan Reduksi Dimensi PCA pada Indikator Kesehatan dan Sosial

2025-08-13T18:28:16+07:00

Public health in West Java faces complex challenges, including disparities in healthcare access, malnutrition, and socio-economic inequalities across districts. These conditions require data-driven analysis to identify patterns of disparity and provide evidence-based guidance for policy intervention. This study aims to cluster districts/cities in West Java based on health and social indicators using Principal Component Analysis (PCA) for dimensionality reduction, followed by K-Means and K-Medoids algorithms for clustering. Data from 27 districts/cities during 2019–2024 were analyzed after standardization. PCA extracted two principal components explaining 61.4% of the total variance. Scree plot and silhouette results indicated three optimal clusters. Comparative analysis revealed that the average silhouette score of K-Means was 0.31, while K-Medoids achieved a higher score of 0.34, suggesting more stable and robust partitioning against outliers. In 2024, Cluster 1 consisted of regions with adequate healthcare facilities and lower prevalence of underweight children; Cluster 2 grouped regions with limited health infrastructure and higher malnutrition problems, while Cluster 3 showed intermediate conditions. Therefore, K-Medoids outperformed K-Means by producing more consistent clustering across years. These findings offer practical recommendations: Cluster 2 should be prioritized for interventions such as improving primary healthcare access and nutrition programs, Cluster 1 requires maintenance of service quality, and Cluster 3 should be targeted for gradual reinforcement.

Pemanfaatan Artificial Intelligence Dalam Implementasi Chatbot Helpdesk untuk Mendukung Layanan TIK Publik pada Instansi Pemerintahan

2025-08-26T15:31:21+07:00

Public services in many government institutions are still traditional, characterized by slow processes, bureaucracy, and limited working hours. This condition makes it difficult for the public to obtain information quickly and accurately. This study aims to analyze the utilization of Artificial Intelligence (AI) in the form of chatbots as a digital innovation to improve the quality, efficiency, and accessibility of public services. The research method used is a qualitative approach through literature review, observation, and interviews with ICT Public Service staff or employees at government agency, as well as users from the community. The research findings, as presented in Table 1 in Discussion sub-chapter 3.4, show that chatbots are able to provide instant and consistent responses, operate more efficiently (saving human resources and time), respond within one minute (real-time, 24/7), support automation, and reduce the workload of government officials through interactive services available around the clock. However, key challenges remain, including limitations in natural language understanding, data security issues, and user resistance to new technologies. With appropriate development strategies and attention to digital literacy among citizens, AI-based chatbots have the potential to become an effective solution in supporting the transformation of public services toward smart, transparent, and citizen-oriented governance.

Penerapan Algoritma Naive Bayes Untuk Sistem Klasifikasi Status Gizi Bayi Balita

2025-04-24T10:39:21+07:00

Infants and toddlers are in a critical period of rapid growth and development, often referred to as the "golden age." During this stage, regular nutritional assessments are essential to monitor health status and detect potential nutritional problems early. This study aims to classify the nutritional status of infants and toddlers using the Naïve Bayes algorithm, a probabilistic classification method based on Bayes' theorem with a strong assumption of attribute independence. The main attributes used in the classification system include age, weight, and height. The dataset consists of 700 records of infants and toddlers collected from previous observations. The results show that the Naïve Bayes algorithm can be effectively implemented for nutritional status classification, achieving a system accuracy of 88.14%. This indicates that the method performs well and has the potential to be utilized in decision support systems for child health monitoring.

Sistem Pendukung Keputusan Penentuan Kelayakan Pemberian Kredit UMKM Menggunakan Metode AHP dan Weighted Product

2025-07-11T20:39:44+07:00

Microfinance institutions often conduct manual evaluations of creditworthiness for UMKM, resulting in subjective and inconsistent decisions. This study aims to develop a decision support system for determining creditworthiness for UMKM using the Analytical Hierarchy Process (AHP) and Weighted Product (WP) methods. AHP is used to evaluate the relative importance of criteria through pairwise comparisons. This process generates weights for each criterion with a consistency matrix as a validation tool. The results of the weighting are then used in the WP method, which calculates the final score for each prospective borrower by multiplying the performance value against the criteria and weights. The case study in this research is at the Koperasi Simpan Pinjam (KSP) Makmur Jaya. The results obtained show that the system is capable of producing objective alternative rankings, where prospective borrowers with the highest VI values are considered the most eligible to receive credit. After testing, the system also demonstrated consistency with manual calculations. Overall, this study shows that the combination of the Analytical Hierarchy Process (AHP) and Weighted Product (WP) can be effectively applied in multi-criteria decision-making in the microfinance sector.

Implementasi Metode Prototype Pada Sistem Informasi Penjualan Berbasis Web

2025-07-25T22:37:52+07:00

In today's digital era, the use of technology has become a primary need in supporting business activities, including in the micro, small, and medium enterprise sector, MSMEs are required to be more digitally literate in order to compete and win the competition in the business world. MertaBeauty is an MSME engaged in the sale of cosmetic and skincare products, but still uses manual methods in managing sales, stock, and reports, which hinders operational efficiency in running its business has not maximized the use of digital which makes the transaction process slow and allows errors in recording. This study aims to build a web-based sales information system to simplify the transaction process, product management, and report preparation. The method used is prototyping, with the final result being a web-based system that supports the management of products, transactions, stock, promotions, and sales reports. The implementation of this system is expected to increase operational effectiveness, as well as support MertaBeauty's digital marketing strategy more optimally. The prototyping method is used because it is a flexible, efficient, and user-centered way to develop products. It helps ignite the fire, minimize risks, and produce products that are more in line with market and user needs.

Implementasi Entreprises Resource Planning Berbasis Web dan Mobile Menerapkan Metode SCRUM

2025-07-23T20:49:31+07:00

Companies operating in the property sector have highly complex business processes involving multiple divisions, such as engineering, marketing, legal, and finance. However, many property companies still manage their data manually and in a fragmented manner, leading to various risks such as data entry errors, communication failures, and other inefficiencies. This study aims to implement an integrated web- and mobile-based Enterprise Resource Planning (ERP) system to support and streamline business processes in a property company, making them more efficient and effective. The development methodology used is Agile, with data collected through interviews, observation, and documentation studies. The system was developed using web and mobile technologies to provide users with flexible access. The implementation results show that the developed ERP system is capable of supporting and improving business processes in the property sector, facilitating real-time data tracking, and increasing operational efficiency. With this system, the company no longer needs to rely on manual data recording and can improve the accuracy of decision-making. This research demonstrates that a digitally based ERP system can be a strategic and effective solution for property companies in facing the challenges of the modern era.