Agriculture is a key contributor to Indonesia's economic growth, with tubers representing the second most important food crop. Despite their significance, the export value of Indonesia’s tuber crops has not yet reached its full potential given the decline in the value of tuber exports since 2021. One of the contributing factors is the restricted range of export market options. This study aims to analyze export trade patterns to identify the most high-potential markets for Indonesian tuber commodities. Clustering analysis is used as a key method to identify market locations by grouping countries based on similar trade characteristics. Clustering was conducted using the Gaussian Mixture Model (GMM), which enhanced by Particle Swarm Optimization (PSO) and evaluated by silhouette score and DBI. The dataset is collected from Indonesia’s Central Bureau of Statistics from 2019 to 2023, focusing on 5 kinds of tuber exports with total of 455 entries and 8 columns. Using the AIC/BIC method, the optimal number of clusters obtained is 2 which are low market opportunities (cluster 0) and high market oppurtunities (cluster 1). Results showed that the GMM model without optimization has silhouette score of 0.7602 and DBI of 0.8398, while the GMM+PSO model achieved an improved silhouette score of 0.8884 and DBI of 0.5584. Both score are categorized as strong structure but, GMM+PSO has higher silhouette score and lower DBI score, demonstrating the effectiveness of PSO in enhancing the clustering model’s performance. The key potential markets for Indonesian tuber exports are primarily concentrated in Asia, including countries such as China, Malaysia, Thailand, Vietnam, Hong Kong, and United States.

This research focuses on accurately classifying pneumonia in children under the age of 5 using X-ray images, considering the challenge of an imbalanced dataset. A modified VGG-16 CNN architecture is evaluated for pneumonia classification in Chest X-Ray Images. The study compares testing results with and without data augmentation techniques and explores the potential application of the model in an Android-based machine learning system for pneumonia diagnosis assistance. Using a dataset of 5,856 Chest X-Ray images categorized as normal or pneumonia, obtained from Kaggle, the research conducts two test scenarios: one without data augmentation and another with data augmentation techniques. The modified VGG-16 CNN algorithm's performance is evaluated using the accuracy metric. The results highlight the effectiveness of data augmentation in improving pneumonia classification accuracy. The augmented tests outperform the non-augmented ones, achieving an impressive 92% accuracy, indicating a significant 15% improvement over the non-augmented scenario. This improvement underscores the efficacy of data augmentation techniques in enhancing the CNN's ability to accurately classify pneumonia, particularly when faced with an imbalanced dataset. Furthermore, the research explores the potential integration of the trained model into an Android-based machine learning system for pneumonia diagnosis assistance. This integration would enable doctors to analyze X-ray images and identify potential pneumonia cases in patients. The integration of advanced machine learning systems in healthcare holds promise for improving patient care and the accuracy of pneumonia diagnoses. In summary, this research contributes to the accurate classification of pneumonia in children under 5 years old using X-ray images. It emphasizes the efficacy of data augmentation techniques in enhancing classification accuracy and explores the practical application of an Android-based machine learning system for pneumonia diagnosis assistance. These findings underscore the importance of advanced machine learning systems in healthcare and their potential to improve pneumonia diagnosis accuracy and enhance patient care.

Currently, the government has implemented performance digitization through information systems that are published through official channels owned by the government, one of which is the government of Kabupaten Malang. The objective of this research was to assess or gauge the measurement and test variables and indicators that affect the quality of the Kabupaten Malang government website with the link www.malangkab.go.id/mlg The problem is, not many governments have launched applications paying attention to the factors that influence user satisfaction so that the government has not been able to prioritize repairs and optimize website performance to meet constituent needs that continue to grow in the digital era. This research employs a survey to identify the causal elements that impact the factors contributing to user satisfaction on the website. The causal factors include website service quality, information quality, and usability quality in user satisfaction. Respondents used in this study were website operators for regional apparatus in Kabupaten Malang, consisting of 81 respondents who met the requirements. In obtaining valid and reliable data, multiple linear regression and hypothesis testing were carried out. There are 4 multiple linear regressions that are carried out, namely, multicollinearity test, autocorrelation test, heteroscedasticity test, and normality test. The results of the influence of service quality, and information quality on user satisfaction through usability quality are 5 models that have a significant influence, that is Service Quality to Usability Quality, Information Quality to Usability Quality, Service Quality to User Satisfaction, Information Quality to User Satisfaction, and Usability Quality to User Satisfaction.

Nowadays, accessibility to academic papers has been significantly improved with electric publications on the internet, where open access has become common. At the same time, it has increased workloads in literature surveys for researchers who usually manually download PDF files and check their contents. To solve this drawback, we have proposed a reference paper collection system using a web scraping technology and natural language models. However, our previous system often finds a limited number of relevant reference papers after taking long time, since it relies on one paper search website and runs on a single thread at a multi-core CPU. In this paper, we present an improved reference paper collection system with three enhancements to solve them: (1) integrating the APIs from multiple paper search web sites, namely, the bulk search endpoint in the Semantic Scholar API, the article search endpoint in the DOAJ API, and the search and fetch endpoint in the PubMed API to retrieve article metadata, (2) running the program on multiple threads for multi-core CPU, and (3) implementing Dynamic URL Redirection, Regex-based URL Parsing, and HTML Scraping with URL Extraction for fast checking of PDF file accessibility, along with sentence embedding to assess relevance based on semantic similarity. For evaluations, we compare the number of obtained reference papers and the response time between the proposal, our previous work, and common literature search tools in five reference paper queries. The results show that the proposal increases the number of relevant reference papers by 64.38% and reduces the time by 59.78% on average compared to our previous work, while outperforming common literature search tools in reference papers. Thus, the effectiveness of the proposed system has been demonstrated in our experiments.

The increasing demand for data-driven decision making to maintain the innovations and competitiveness of organizations highlights the need for data science educations across academia and industry. At its core is a solid understanding of statistics, which is necessary for conducting a thorough analysis of data and deriving valuable insights. Unfortunately, conventional statistics learning often lacks practice in real-world applications using computer programs, causing a separation between conceptual knowledge of statistics equations and their hands-on skills. Integrating statistics learning into Python programming can convey an effective solution for this problem, where it has become essential in data science implementations, with extensive and versatile libraries. In this paper, we present a self-learning method for fundamental statistics through Python programming for data science studies. Unlike conventional approaches, our method integrates three types of interactive problems—element fill-in-blank problem (EFP), grammar-concept understanding problem (GUP), and value trace problem (VTP)—in the Programming Learning Assistant System (PLAS). This combination allows students to write code, understand concepts, and trace the output value while obtaining instant feedback so that they can improve retention, knowledge, and practical skills in learning statistics using Python programming. For evaluations, we generated 22 instances using source codes for fundamental statistics topics, and assigned them to 40 first-year undergraduate students at UPN Veteran Jawa Timur, Indonesia. Statistics analytical methods were utilized to analyze the student learning performances. The results show that a significant correlation ($\rho <0.05$) exists between the students who solved our proposal and those who did not. The results confirm that it can effectively assist students in learning fundamental statistics self-learning using Python programming for data science implementations.

Stock price data or similar time series data can be used to carry out forecasting processes using past data. The method that can be used is like a neural network, one type of neural network that is used is the Recurrent Neural Network. When using the Recurrent Neural Network (RNN) method, we need to determine the appropriate parameters in order to get the best forecasting results. It takes experience or . In this study, this problem can be solved using optimization algorithms, such as Genetic Algorithms. With genetic algorithms, neural networks can be trained to get the best objective function. So that after implementing the RNN which was optimized using the Genetic Algorithm on stock time series data, when the trial was carried out without optimization the Genetic Algorithm got an RMSE value of 0.108, after being combined using the genetic algorithm it got an RMSE value of 0.106.