This study evaluates the performance of three machine learning models—Random Forest, Support Vector Machine (SVM), and Logistic Regression—in predicting heart disease using the "Heart Disease UCI" dataset from Kaggle. The models were assessed based on accuracy, precision, recall, and F1-score, both with and without feature selection techniques such as Chi-Square and Mutual Information.Without feature selection, Random Forest achieved the highest performance with an accuracy of 89.7%, followed by SVM with 87.0%, and Logistic Regression with 84.2%. Using Mutual Information for feature selection, Random Forest achieved an accuracy of 85.3%, SVM 87.0%, and Logistic Regression 82.6%. With Chi-Square feature selection, Random Forest and Logistic Regression both showed an accuracy of 83.2%, while SVM achieved 82.6%.The results indicate that Random Forest consistently performs well across different scenarios, making it a robust choice for heart disease prediction. Feature selection did not significantly enhance model performance, suggesting that the initial features in the dataset are already highly relevant. These findings highlight the potential of machine learning, especially Random Forest, in aiding clinical diagnosis of heart disease. Further research is needed to validate these models on larger, more diverse datasets and to explore advanced feature selection techniques for improved model performance.

In the digital era, the accessibility of vast textual data, including the Quran, has facilitated broader comprehension of its teachings. This study analyzes the emotions in the English translation of Surah Yusuf using the NRC Emotion Lexicon. The findings show that trust is the most dominant emotion (22.89%), followed by joy (15.66%), anticipation (13.25%), sadness (12.05%), fear (10.84%), anger (9.64%), surprise (8.43%), and disgust (7.23%). These results confirm the text's diverse emotional expressions and the effectiveness of the lexicon-based method. The research aligns with the initial goals and highlights the potential of emotion analysis in understanding religious texts. Future research can expand the analysis to more verses and use machine learning for improved accuracy. This study aids scholars and students in exploring the Quran's emotional and spiritual dimensions and can be adapted to other texts for broader applications.

In an increasingly competitive era, it is crucial for car dealers and retailers to address the challenges of accurately determining the prices of used cars. To tackle these challenges, this study implements Machine Learning models to predict used car prices accurately. By applying the Artificial Neural Network (ANN) and Random Forest Regression algorithms, this research aims to evaluate the performance of these methods in predicting used car prices. The used car price data was obtained from the Kaggle repository, consisting of 14,657 data entries that provide comprehensive information about used cars. The analysis focuses on six main columns, including Brand, Model, Variant, Year, and Mileage, to estimate used car prices. Model evaluation was conducted using Mean Absolute Error (MAE) as the primary metric. The results show that the ANN model achieved a lower MAE (0.035) compared to the Random Forest Regression (0.047), indicating better performance in predicting used car prices. These findings demonstrate the effectiveness of ANN in handling data complexity and the non-linear relationships between variables involved in forecasting used car prices. Additionally, this contributes to the implementation of more accurate used car price predictions, enabling automotive companies to improve operational efficiency and provide greater benefits to the community.

The industrial revolution has significantly increased greenhouse gas emissions, leading to global warming and climate change, which burdens producers worldwide. In Indonesia, PT United Tractors Tbk, as a major player in the heavy equipment industry, faces challenges in the environmental reporting process due to the use of separate data collection methods between digital and non-digital (manual paper- based), which results in a lack of proper integration among aspects, making it impossible for environmental reporting to be done in real- time and causing difficulties in tracking the reporting data. This research aims to design an intuitive user interface (UI) and a seamless, satisfying user experience (UX) for a mobile-based environmental reporting application using the User-Centered Design (UCD) method, focusing on integrating all aspects of environmental reporting —including water, hazardous waste, non hazardous waste, and air—into a practical mobile platform for the company's environmental staff. The System Usability Scale (SUS) method was then employed to evaluate user satisfaction and acceptance of the developed application. The SUS results from 12 prospective users among the company's environmental staff showed an average score of 89.4, with the lowest score being 82.5 and the highest score being 100, indicating an excellent score and demonstrating that the UI/UX design is highly satisfactory and well-received by users.

Bladder cancer is one type of tumor that frequently occurs in the urinary system, and early diagnosis is essential to improve the prognosis and survival of patients. The study aims to develop a Convolutional Neural Network (CNN) model for bladder tissue lesion classification from endoscopic images. This study uses a dataset consisting of 1754 images, which are divided into four classes: High-Grade Cancer (HGC), Low-Grade Cancer (LGC), Non-Specific Tissue (NST), and Non-Tumorous Lesion (NTL). The proposed CNN model showed a validation accuracy of 96.29%, with high recall, precision, and F1-score in most classes. The results show that CNN-based automated methods can improve efficiency and accuracy in the early diagnosis of bladder cancer, reduce manual visual interpretation errors, and improve the quality of patient care. This study suggests increasing the training data, especially for the NTL class, and applying more complex model architecture to better results.

Drug users or abusers are people who use narcotics or psychotropic drugs without supervision or medical indication from a doctor. Before undergoing rehabilitation, drug users must first undergo an examination to determine their level of drug dependence so that they can receive medical treatment according to their level of drug dependence. Determining the level of drug dependence requires a technique that can provide labels or categories of data for drug users based on the user's condition or influential criteria. This study applies the Naïve Bayes Classifier method to a system to determine the level of drug dependence. This study uses medical record data from 220 drug users. The user's medical record data is processed using data mining stages consisting of data selection, data cleaning, data transformation, and division of training and test data to produce 120 training data and 100 test data. The results of the Naive Bayes Classifier method calculation resulted in 29 users having a trial level of dependence (mild), 42 identified as having a regular level of dependence (moderate), and 29 others as users with a severe level of dependence. The confusion matrix testing was very accurate, namely, 94% accuracy, 95% precision value, and 92% recall. Meanwhile, the system that has been built can run very well. Based on the results of the research that has been conducted, this research can contribute to determining the level of dependence of drug addicts objectively so that related parties can provide rehabilitation or appropriate treatment to drug addicts.

In the rapidly evolving landscape of digital technology, the proliferation of interconnected systems has brought unprecedented opportunities and challenges. Among these challenges, the escalating frequency and sophistication of cyberattacks pose significant threats to individuals, organizations, and nations. In response, the fusion of Cybersecurity and Artificial Intelligence (AI) has emerged as a pivotal paradigm, offering proactive, intelligent, and adaptable defense mechanisms. This research explores the transformative impacts of AI-powered security on cybersecurity, demonstrating how AI techniques, including machine learning, natural language processing, and anomaly detection, fortify digital infrastructures. By analyzing vast volumes of data at speeds beyond human capacity, AI-driven cybersecurity systems can identify subtle patterns indicative of potential threats, allowing for early detection and prevention. The exploration consolidates existing studies, highlighting the trends and gaps that this research addresses. Expanded results and discussions provide a detailed analysis of the practical benefits and challenges of AI applications in cybersecurity, including case studies that offer concrete evidence of AI's impact. Novel contributions are emphasized through comparisons with other studies, showcasing improvements in accuracy, precision, recall, and F-score metrics, which demonstrate the effectiveness of AI in enhancing cybersecurity measures. The synergy between AI and human expertise is explored, highlighting how AI-driven tools augment human analysts' capabilities. Ethical considerations and the "black box" nature of AI algorithms are addressed, advocating for transparent and interpretable AI models to foster trust and collaboration between man and machine. The challenges posed by adversarial AI, where threat actors exploit AI system vulnerabilities, are examined. Strategies for building robust AI security mechanisms, including adversarial training, model diversification, and advanced threat modeling, are discussed. The research also emphasizes a holistic approach that combines AI-driven automation with human intuition and domain knowledge. As AI continues to rapidly evolve, a proactive and dynamic cybersecurity posture can be established, bolstering defenses, mitigating risks, and ensuring the integrity of our increasingly interconnected digital world.

Traditional fishing techniques frequently lack efficiency and optimization, resulting in fishermen obtaining unsatisfactory yields. This study presents a novel approach by incorporating Geographic Information System (GIS) technology, notably utilizing Leaflet, to improve fishing techniques. The suggested system incorporates a LoRa node tool that logs the journeys of fishermen, offering comprehensive itineraries and data on the distribution of fish and unfavorable weather conditions. Notable outcomes were attained by employing the haversine approach to compute distances between the LoRa Gateway and different data points. The approach exhibited a negligible error margin of 0.157% in contrast to the calculations performed in Excel. In addition, the GPS accuracy testing produced remarkable results, with latitude and longitude errors of 0.000116% and 0.000002%, respectively. The LoRa system demonstrated a range of RSSI performance, with values ranging from -57 dBM at 50 meters to -121 dBM at 1500 meters. This range of performance guarantees dependable transmission of data over significant distances. The findings underscore the GIS-based strategy's efficacy in enhancing the effectiveness and precision of conventional fishing methods, presenting a promising technical improvement for the fishing sector.

One of the deadly diseases that attacks many women is breast cancer. It was recorded that breast cancer cases in 2020 were 2.3 million, with deaths accounting for 29% of these cases. The BSE technique is an easy way of early identification of breast cancer that can be done independently. However, this technique often goes wrong when practiced, making it ineffective. An early breast cancer detection system is proposed to make it easier for women to carry out early identification independently. Detection is carried out based on the measured temperature of the breast surface. The temperature difference at each point is a reference for the potential for breast cancer. This system was built in a bra and tested with a mannequin as a simulator subject. The MLX90614 temperature sensor, as the primary sensor, succeeded in measuring the surface temperature of the dummy with 99% accuracy. Final testing of the proposed system can also differentiate the temperature differences in each zone.

Most Indonesian social media users under the age of 25 use various words, which are now often referred to as slang, including abbreviations in communicating. Not only causes, but this variation also poses challenges for the natural language processing of Indonesian. The previous researchers tried to improve the Recurrent Neural Network to correct errors at the character level with an accuracy of 83.76%. This study aims to normalize abbreviated words in Indonesian into complete words using a Recurrent Neural Network in the form of Bidirected Long Short-Term Memory and Gated Recurrent Unit. The dataset is built with several weight confgurations from 3-Gram to 6-Gram consisting of words without vowels and complete words with vowels. Our model is the frst model in the world that tries to fnd incomplete Indonesian words, which eventually become fully lettered sentences with an accuracy of 97.44%.