Telematika
Not a member yet
319 research outputs found
Sort by
System Usability Scale (SUS) As An Analysis Method For Official Website
Full text linkPurpose: A company expands its product marketing by utilizing information systems. The quality of its information system influences product marketing expansion. Only a few companies have maximized the use of company websites. Usability testing on the company\u27s official website determines the page\u27s usability level.Design/methodology/approach: To address this issue, usability testing is conducted in this research using the System Usability Scale, whose testing involves page users. This usability level measurement technique has its characteristics. This method can obtain the level of usability precisely with user respondents from the page.Findings/result: Website testing was carried out with a structured and accurate SUS questionnaire using ten questions. The SUS score calculation was 80, considered excellent and acceptable to users.Originality/value/state of the art: PT Inka\u27s Information System has never been analyzed during design or development. With this usability test, it is hoped that website users can find out the perceptions and problems experienced by users when interacting with the official websit
Performance Evaluation of Multiple Deep Learning Models for Wine Quality Prediction
No full textResearch utilizing a dataset from the UCI repository evaluated the predictive accuracy of nine machine learning models for wine quality. The models employed include Logistic Regression, K-Nearest Neighbor (KNN), Decision Tree, Support Vector Machine (SVM), Random Forest, XGBoost, LightGBM, CatBoost, and Gradient Boosting. The dataset comprises 1,599 samples with 12 chemical parameters. Data preprocessing, including oversampling, normalization, standardization, and seeding, was performed to enhance model performance.The study\u27s findings indicate that the models with the highest accuracy values were LightGBM (87.80%), CatBoost (86.60%), and Random Forest (85.70%). A voting classifier combining these three models achieved an accuracy of 87.29%. Further analysis using a confusion matrix demonstrated that this combined model effectively predicts the "Good" and "Not Good" classes.In conclusion, the combination of LightGBM, CatBoost, and Random Forest models proves to be an effective approach for predicting wine quality based on chemical parameters, with an accuracy value of 87.29%
Tahsin Tracker as an Effort to Improve the Services of the Islamic Studies Development Institute for Lecturers and Education Personnel at Ahmad Dahlan University
Full text linkPurpose: This research aims to produce a Tahsin Tracker application as a tool to overcome several problems in managing tahsin guidance organized by the Islamic Studies and Development Institute (LPSI) UAD for lecturers and education staff.Design/method/approach: The method used in system development by applying general software development methods. The stages adopted include data collection, needs analysis (user and system), design (process, database and interface), implementation (database and program, and system testing.Results: A Mobile Web-based Tahsin Tracker application has been produced and has been tested on potential users using black box and SUS testing. Black box testing is 100% valid. The SUS test obtained a score of 79. Based on the SUS score of 79, the application is in the Acceptance Range: ACCEPTABLE, Grade Scale: C, and Adjective Rating: EXCELLENT category.Originality/state of the art: Based on previous research, as well as the results of the development of existing tahsin applications, this research adopts and complements existing deficiencies and produces new assessments of the features and usability in related units, because it replaces the manual method of book record basis, and becomes the main alternative for developing institutional service media in a better direction.
Integrating Multiple Machine Learning Models to Predict Heart Failure Risk
Full text linkThe research aims to create and evaluate machine learning models for the prognosis of heart failure based on patient medical information. Various predictive models have been created employing algorithms like logistic regression, decision trees, random forests, K-nearest neighbors, naive Bayes, support vector machines (SVMs), neural networks, and ensemble voting classifiers. The dataset utilized comprises diverse clinical characteristics from patients diagnosed with heart failure. The data underwent division into training and testing sets in an 80:20 ratio. Metrics including accuracy, Cross Validation Score, and ROC_AUC Score score were used to assess the models\u27 performance. The findings reveal that the Voting Classifier, amalgamating the Logistic Regression and Support Vector Classifier models, demonstrated superior performance with an accuracy of 88.04%, a cross-validation score of 91.01%, and a ROC_AUC score of 88.00%. Further scrutiny suggested that blood pressure and cholesterol levels serve as substantial indicators of heart failure. This study presents a notable advancement in the utilization of machine learning models for heart failure prediction by scrutinizing diverse algorithms and pinpointing the most pertinent clinical characteristics. These outcomes hint at the potential for the development of machine learning-driven clinical tools to facilitate early detection and enhance medical interventions
The Application of Artificial Intelligence in Waste Classification as an Effort In Plastic Waste Management
Full text linkPurpose: Sorting waste before it is deposited in the Final Disposal Site (TPA) is crucial to reduce the increasing amount of waste accumulation each year. This issue can be addressed by implementing machines capable of automatically sorting waste.Design/methodology/approach: This research is quantitative and utilizes secondary data, namely image data of various types of waste. The images will be classified into organic and inorganic waste using a deep learning model. The measurement conducted involves assessing the accuracy of the designed deep learning model in classifying waste images into appropriate categories.Fondings/results: Based on the available dataset, waste identification will be performed, including food waste, paper, wood, leaves, electronic waste, metal, plastic, and bottles. The overall accuracy of the model is 94.42%, indicating that the model correctly classifies 94.42% of waste samples.Originality/value/state of the art: This research can classify 8 types of waste classes successfully using deep learning
Sensitivity Comparison of AHP with The Combination of AHP and SAW for Facial Wash Recommendation System based on Skin Type
Full text linkPurpose: This research aims to design a facial wash recommendation system based on all skin types, namely normal, dry, oily, combination, and sensitive. This is to tackle the limitation of previous systems that were developed based on limited skin types which are normal, dry, and oily using Promethee II, Fuzzy Logic, and SAW methods.Design/methodology/approach: This research uses the Analytic Hierarchy Process (AHP) method and a combination of AHP and Simple Additive Weighting (SAW) to consider the importance values of each criterion. Four criteria data are used, namely price, rating, content, and availability, along with 70 alternative data of facial wash products.Finding/Result: Sensitivity testing was conducted on both methods, and the combination of AHP and SAW produced a higher sensitivity percentage, which is 67.51%, whereas the AHP method provided a lower sensitivity percentage of 59.26%.Originality/state of the art: The combination of AHP and SAW is an innovation in designing a facial wash recommendation system, and the research results demonstrate that the combination of AHP and SAW is a superior method for recommending facial wash products
The Implementation of Color Feature Extraction and Gray Level Co-occurrence Matrix Combination in K-Nearest Neighbor Classification Method for Tomato Leaf Disease Identification
Full text linkPurpose: Tomato plants are quite important commodities in Indonesia. With a complete and good content of substances, tomatoes become a product that is widely consumed by the public. However, much of the decline in crop production is caused by plant disruptive organisms such as viruses and bacteria. Early identification of plant diseases is expected to prevent the spread of diseases caused by these organisms.Design/methodology/approach: In this study the data used in machine training are data from kaggle sites. This study uses the K-Nearest Neighbor classification method with a combination method of extracting feature on RGB, HSV and GLCM images to obtain the best accuracy value.Findings/Results: Based on the test results among the combination methods of feature extraction in the process of identifying tomato leaf diseases which are classified into 7, namely testing units of RGB, HSV, GLCM followed by a combination of RGB HSV, RGB GLCM, HSV GLCM, and RGB HSV GLCM methods obtained a comparison value of 71.5%, 72.9%, 79%, 82.5%, 90.6%, 87.4% and 87.7%. Based on these data, it was concluded that with the combination of the RGB GLCM method obtained the best accuracy value in the identification of tomato leaf disease with an accuracy rate of 90.6%.Originality/value/state of the art: The use of the K-Nearest Neighbor classification method in this study combines the collection of selected characteristics so as to get a comparison of 7 combination groups between RGB, HSV, and GLCM
Design of a Generative AI Image Similarity Test Application and Handmade Images Using Deep Learning Methods
Full text linkPurpose: The aim of this research is to develop a classification model using the Transformer approach, specifically the BEiT architecture, to differentiate between handmade images and AI Generative Art. The objective is to ensure the authenticity of art and address ethical and legal concerns related to AI Generative Art.Design/methodology/approach: The study utilizes the BEiT architecture within the Transformer approach to create a classification model. The training process uses Bidirectional Encoder representation from Image Transformers (BEiT) to improve image classification. The primary datasets are collected through a Python image scraper program. The BEiT workflow includes Pre-training, Masking, Inpainting, and Interface Design with Gradio.Findings/result: The Transformer model, using the BEiT architecture, achieves 96.34% accuracy and 0.0921 loss in differentiating handmade images and AI Generative Art. The model demonstrates a balanced precision and recall in each category, outperforming previous methods such as Convolutional Neural Network (CNN) and VGG16. The language used is clear, objective, and value-neutral, with a formal register and precise word choice. No changes in content were made. The Gradio interface was used to successfully test the model.Originality/value/state of the art: The research presents a state-of-the-art classification model that uses the Transformer approach, specifically the BEiT architecture, to differentiate between handmade and AI Generative Art images. The research presents a state-of-the-art classification model that uses the Transformer approach, specifically the BEiT architecture, to differentiate between handmade and AI Generative Art images. The text adheres to conventional structure and formatting features, including consistent citation and footnote style. The sentences and paragraphs create a logical flow of information with causal connections between statements. The text is free from grammatical errors, spelling mistakes, and punctuation errors. Additionally, the research is enhanced by the innovative approach to data collection using a Python image scraper program
Klasifikasi Penyakit Gangguan Jiwa menggunakan Metode Logika Fuzzy
Full text linkPurpose: This research aims to facilitate psychologists in handling individuals with mental disorders by categorizing them based on their symptoms and conditions using fuzzy logic, which mimics the functioning of the human brain.Design/methodology/approach: The categorization is performed by applying Mamdani fuzzy logic, designed in consultation with psychology experts. Ten initial symptoms each have parameters (Mild, Moderate, and Severe) as input variables, and the output variable involves mental health disorders such as Schizophrenia, Bipolar disorder, Eating disorders, and Anxiety. The fuzzy process employs the Mamdani method with IF-THEN rules and AND operators. The implementation of Mamdani fuzzy logic achieves adequate accuracy in classifying individuals with mental disorders, providing a strong foundation for a more targeted psychological approach. In the context of accuracy, fuzzification analysis for each health disorder can offer further insights.Findings/result: Results of the study for Schizophrenia, for instance, show a fuzzy diagram membership of approximately 0.4, indicating a potentially high level of thought impairment and interpersonal skills. Weighting for low, medium, and high is then assessed to categorize patients. A similar process is undertaken for Bipolar disorder, with special attention to the middle value and the strong relationship between two input values. Regarding mental illness, membership analysis indicates an increasing level of membership corresponding to condition groups, suggesting compatibility with existing rules.Originality/value/state of the art: These findings reinforce the Mamdani fuzzy logic implementation as a reliable approach in classifying individuals with mental disorders, with the potential to enhance psychological diagnosis and interventions more effectivel
Application Random Forest Method for Sentiment Analysis in Jamsostek Mobile Review
Full text linkPurpose: This study aims to monitor the service quality of JMO applications from time to time by classifying JMO user reviews into the class of positive, neutral, and negative sentiments.Design/methodology/approach : The method used in this study is the random forest classification method. Data processing in this study uses feature extraction, TF-IDF and labeling with the lexicon-based method.Findings/result: Based on the research results, it was found that the highest frequency of classification was the positive class with 17571 reviews compared to the neutral class with 8701 reviews and the negative class with 3876 reviews with an accuracy evaluation value of 93%, precision 88%, recall 93%, and f1-score 90%.Originality/value/state of the art:This study uses 150737 reviews that have been pre-processed using the random forest method and TF-IDF and lexicon-based feature extraction