Telematika
Not a member yet
319 research outputs found
Sort by
Performance Analysis of SVM Kernels in Sentiment Classification on Indonesian Local Skincare Dataset
Full text linkPurpose: Sentiment analysis is an important aspect of understanding consumers\u27 views on products, especially in the growing skincare industry. This study aims to compare the accuracy and effectiveness of various kernels in the Support Vector Machine (SVM) algorithm, including linear, polynomial (poly), and radial basis function (RBF) kernels, in predicting three types of sentiment: positive, neutral, and negative based on reviews of local Indonesian skincare products.Design/methodology/approach: The dataset used includes consumer reviews classified by rating, which are then processed using Term Frequency-Inverse Document Frequency (TF-IDF) technique for feature extraction.Findings/result: The evaluation results show that the RBF kernel achieves the highest accuracy of 74.78%, followed by the linear kernel with 74.51% accuracy, and the polynomial kernel with 74.10% accuracy. Although the difference between the three kernels is not significant, the RBF kernel excels in positive sentiment classification, while all three kernels struggle in predicting neutral and negative classes.Originality/value/state of the art: These findings make an important contribution to the development of effective sentiment analysis methods, especially in the context of datasets with high class imbalance. To handle class imbalance, techniques such as oversampling smaller classes or using cost-sensitive learning techniques to give more weight to negative and neutral classes can be used.
Strategic Planning of IT/IS Using Ward and Peppard (PT Wahid Bangun Semesta Yogyakarta)
Full text linkPurpose: This research aims to design an IS/IT strategy for PT Wahid Bangun Semesta to propose prioritized information systems that align with the company\u27s existing challenges and to provide guidelines for application development. Design/methodology/approach: This study focuses on designing an IS/IT strategy using the Ward and Peppard methodology. Findings/result: Based on the findings, PT Wahid Bangun Semesta Yogyakarta is advised to implement an integrated information system to support operations and enhance competitiveness. Fingerprint systems are recommended to reduce attendance misuse. The application portfolio is designed to strengthen business processes according to identified needs. The IT strategy includes LAN and wireless networks at the head office, as well as USB modems and web-based applications at branch offices for flexible information access. Establishing an IT division is also recommended for more focused management, enabling IS/IT investments to positively contribute to company growth. Originality/value/state of the art: This study designs an IS/IT strategy for PT Wahid Bangun Semesta Yogyakarta to enhance operations and competitiveness through integrated information solutions. Proposed solutions include implementing an Integrated Information System, a fingerprint machine to prevent attendance misuse, LAN and wireless networks at the head office, broadband access at branch offices, and web-based applications.Establishing an IT division is recommended to manage IS/IT resources and ensure successful investments that support the company’s growth.
The Improving Cross-Project Software Defect Prediction with CORAL-Based Domain Adaptation and Ensemble Learning
No full textAbstract—This study presents a cross-project software defect prediction (CSDP) framework combining feature harmonization, CORAL-based domain adaptation, SMOTE balancing, PCA reduction, and ensemble classifiers: Random Forest, Logistic Regression, XGBoost, AdaBoost, and VotingClassifier. Evaluations on five AEEEM datasets (JDT, EQ, PDE, Lucene, Mylyn) in both single-source and multi-source settings show consistent improvements over baseline methods. While not outperforming deep learning models, the approach remains practical and interpretable for real-world CSDP tasks
Detection Of CT Kidney Disease Using GLCM Feature Extraction and Kernel Extreme Learning Machine (KELM) Classification Method
Full text linkKidney disease includes a variety of disorders affecting renal function, such as kidney cysts, tumors, and stones. If left untreated, these conditions can progress to chronic kidney disease, posing significant health risks and potentially leading to mortality. This study aims to classify kidney diseases by using the Gray Level Co-occurrence Matrix (GLCM) for feature extraction and the Kernel Extreme Learning Machine (KELM) as a classification method, with renal CT images as the dataset. The classification process categorizes kidney conditions into four classes: Cyst, Normal, Stone, and Tumor. The dataset consists of 4,232 CT images, with 1,058 images per class, evenly divided into axial and coronal orientations. The study utilizes k-fold cross-validation with k = 5 and k = 10 and implements the Radial Basis Function (RBF) as the kernel function in the KELM model. An iterative tuning of parameters, including the kernel parameter () and the regularization constant (), was conducted to identify the optimal model configuration. The best classification performance was achieved at angle using k = 5, with an accuracy of 97.26%, sensitivity of 97.16%, and specificity of 99.05%. Furthermore, the model demonstrated high computational efficiency, requiring only 6.07 seconds
Analisis Kualitas dan Klasifikasi Jenis Tanah Berbasis Pengolahan Citra: Teknik Image Sharpening dan CNN ResNet untuk pemetaan pemanfaatan Daerah Pesisir
Full text linkTujuan: Penelitian ini bertujuan untuk menganalisis kualitas dan klasifikasi jenis tanah di wilayah pesisir Teluk Kendari dengan menggunakan teknik penajaman gambar dan Convolutional Neural Network (CNN) ResNet152V2, guna mendukung pengelolaan sumber daya wilayah pesisir yang berkelanjutan.Perancangan/metode/pendekatan: Penelitian menggunakan pendekatan pengolahan citra digital dengan tahapan: pengumpulan dataset dari Kaggle dan lapangan, image preprocessing, image sharping, dan klasifikasi menggunakan model CNN ResNet152V2. Dataset terdiri dari 880 gambar dari Kaggle dan 110 gambar dari wilayah Teluk Kendari, dibagi menjadi data latih (80%), uji (10%), dan validasi (10%).Hasil: Model CNN ResNet152V2 berhasil mencapai akurasi klasifikasi sebesar 90.91% dalam mengidentifikasi delapan jenis tanah (Aluvial, Andosol, Entisol, Humus, Inceptisol, Laterit, Kapur, dan Pasir). Teknik penajaman gambar terbukti efektif meningkatkan kualitas citra visual, memperjernih detail tekstur tanah, dan memudahkan proses klasifikasi.Keaslian/ state of the art : Penelitian ini mengintegrasikan teknik penajaman gambar dan CNN ResNet untuk menganalisis tanah pesisir, yang sebelumnya belum banyak dilakukan di Indonesia. Pendekatan ini memberikan kontribusi dalam memahami kondisi tanah di wilayah pesisir dan mendukung strategi pengelolaan berkelanjuta
Application of Fibonacci Pattern for Network QoS (Quality of Service) Management)
Full text linkIn managing network quality of service (QoS), this study uses Fibonacci patterns to optimize delay control and bandwidth allocation. QoS is essential in contemporary network management, especially given the increasing demand for stable and effective data services. This study prioritizes data based on traffic levels using a simulated Fibonacci algorithm. Each priority is assigned a value corresponding to the Fibonacci sequence, which allows allocating resources more in line with the network load. Simulations are performed under normal and overload conditions. The results show that conventional methods, such as round-robin and weighted fair queuing, can improve QoS efficiency with Fibonacci patterns by up to 15%. This improvement mainly concentrates on controlling important data packets such as real-time communication and video streaming and reducing delay. In addition, this technique is better at adjusting to traffic changes. The results show that the Fibonacci pattern can be an innovative method for managing network QoS, especially for complex prioritization requirements. It can be a reliable tool to improve user experience with modern network services if used properly. To find out how Fibonacci patterns relate to future network technologies such as 5G and the Internet of Things, further research is needed
Implementation of Forgy Initialization and K-Means++ Algorithms in the K-Means Clustering Method for Sales Data Analysis of Dazzle Store
Full text linkObjective: To determine the results of K-Means Clustering calculations by applying K-Means++ and Forgy initialization methods in analyzing sales data at Dazzle accessory store, as well as to identify the optimal number of clusters using the silhouette coefficient.Method: This study implements the Forgy initialization and K-Means++ algorithms in the K-Means Clustering method, along with an evaluation of the optimal number of clusters using the silhouette coefficient method.Results: The application of Forgy initialization and K-Means++ successfully improved clustering outcomes more optimally compared to the pure initialization method. The highest silhouette coefficient evaluation score was 0.9232095222373023 for K-Means++ and 0.8822890619277 for Forgy initialization. This result is clearly better than the pure initialization method, which only achieved a score of 0.8816344025002508.State of the Art: This study builds upon previous research. The innovation lies in the implementation of a combination of K-Means Clustering with Forgy initialization and K-Means++ initialization methods
Combination of Deep Neural Network and YuNet for Python-Based Human Lifespan Prediction
Full text linkPurpose: In this research on face detection, many methods face challenges in the accuracy of age prediction due to the complexity of facial features that are influenced by factors such as lighting, expression, and image quality. Therefore, this research focuses on developing more accurate and efficient methods by utilizing Deep Neural Network (DNN) and YuNet. The purpose of this study is to develop a face recognition model in detecting and determining human age automatically using Python with the DNN method to study facial patterns in determining human age precisely and integrate the YuNet library as a lightweight face detection framework that is efficient in the identification process.Design/methodology/approach: In this study, a system was created for predicting human age using the Deep Neural Network method which functions to predict age based on facial patterns in images and the Yunet method as a facial image detector. The stages of this research start from taking pictures, installing python libraries, namely opencv, face detection process, and age detection process.Findings/result: The results of the study show that the DNN and YuNet methods have tested as many as 50 samples in the form of photos of human faces taken at a distance of half a meter, so by using the DNN and YuNet methods, researchers have succeeded in obtaining the age of the human face through the image processing process which can then obtain an accuracy level or Precission of 80% and the accuracy level of success between the prediction value and the actual value given by the system is 80%.Originality/value/state of the art: In this study, the system uses Python tools where in the face detection process using the YuNet method, this method is used because YuNet can directly detect facial features in the image and is lightweight in operation. In terms of DNN prediction, it functions as a method that can predict age based on the results of facial image detection. In this study, a dataset was also used for 50 facial samples that were tested for accuracy using the confussion matrix by looking for precission, recal, and accuracy values.
Comparison of Algorithms for Cyberbullying Detection to Football Player in Social Media
Full text linkPurpose: to compareNaïve Bayes, Support Vector Machine(SVM), and K-Nearest Neighbor(KNN) algorithms for detecting cyberbullying that happen to football player in social media.Design/methodology/approach: In the cyberbullying detection process, the steps involved are data collection, labeling, data preprocessing, feature extraction, modeling, and finally evaluation by comparing the accuracy values of the three methods used.Findings/result: Based on the accuracy values obtained, Naive Bayes emerged as the algorithm with the highest accuracy at 78.6%, followed by Support Vector Machine (SVM) with an accuracy of 77.9%, and K-Nearest Neighbor (KNN) with an accuracy of 65.6%.Originality/value/state of the art: This research discusses the comparison of algorithms for detecting cyberbullying in social media related to football players, an area that has not been addressed by other studies. Additionally, the preprocessing stage and the three algorithms used were also designed and chosen by the researchers themselves.
Perancangan Sistem Pemesanan Menu pada Kedai Teh Berbasis Customer Relationship Management (CRM)
Full text linkTujuan: Penelitian ini bertujuan merancang sistem pemesanan menu mandiri berbasis CRM untuk meningkatkan kinerja operasional dan interaksi pelanggan di Kedai Teh Kaula, sebuah usaha kuliner berskala kecil.
Metodologi: Sistem dikembangkan dengan model waterfall dan kerangka kerja CRM Francis Buttle. Pengumpulan data dilakukan melalui observasi, wawancara, dan telaah pustaka, dilanjutkan dengan pembuatan Use Case Diagram, ERD, DFD, dan wireframe antarmuka.
Hasil: Desain sistem mendukung desain sistem pemesanan mandiri lewat QR code serta fungsi admin untuk pengelolaan pesanan dan data pelanggan. Desain sistem ini mengatasi antrian panjang, meningkatkan akurasi pesanan, dan memperbaiki pengalaman pengguna.
State of the Art: Penelitian ini menunjukkan integrasi praktis strategi CRM pada tahap awal perancangan sistem, memberikan model yang skalabel bagi usaha kecil yang menjalani transformasi digital