Sinkron : jurnal dan penelitian teknik informatika
Not a member yet
1261 research outputs found
Sort by
Strategic Clustering of Poverty Areas in Central Java Using K-Means and Silhouette Evaluation
No full textIndonesia is one of several developing nations that struggle with poverty. Central Java is one of Indonesia's provinces with the third-highest percentage of the country's inadequate. This study aims to explore and improve the application of the K-Means Algorithm in investigating socioeconomic disparities. In this study, the Elbow method is used to determine the optimal number of clusters to overcome the weaknesses in determining the number of clusters in conventional K-Means. Model evaluation using the silhouette coefficient shows the effectiveness of this method approach with a value of 0.504 and several clusters (K = 3), which meets the medium structure category. The Human Development Index (HDI) and Uninhabitable Households (RTLH) were two criteria used to categorize poverty areas using the K-Means Algorithm optimization successfully. According to the clustering results, there were 12 regions in Cluster 0, 2 in Cluster 1, and 21 in Cluster 2. These findings are anticipated to offer the Central Java Provincial Government critical insights, facilitating the development of precise and well-targeted initiatives to address deprivation issues effectively. Furthermore, a more systematic and structured optimization of the K-Means algorithm has the potential to significantly improve both the accuracy and practical relevance of studies on socioeconomic inequality in Central Java Province. This enhanced methodological approach can provide more in-depth results on data-driven regional disparities to reduce these disparities comprehensively
Smart Diagnosis of Coffee Diseases via Web-Based Expert System
No full textIndonesia’s coffee industry faces persistent threats from plant diseases and pests, which significantly impact crop yield and farmer livelihoods. Many smallholder farmers lack access to timely expert guidance, leading to delays in diagnosis and ineffective treatments. This study proposes a web-based expert system designed to assist farmers in diagnosing coffee plant diseases and pests based on observed symptoms. The system integrates a Bayesian Network (BN) to model the probabilistic relationships between symptoms and diseases. It employs a Breadth-First Search (BFS) algorithm to optimize the exploration of symptom-disease associations. Developed using Node.js, Next.js, and MySQL, the system enables users to input their symptoms and receive probabilistic diagnoses along with treatment suggestions. Validation results show over 85% accuracy compared to expert assessments, highlighting the system's reliability and scalability. This research demonstrates that combining probabilistic reasoning and structured graph traversal provides an effective diagnostic tool, especially for underserved rural communities. Furthermore, the system serves as a foundation for future development of intelligent agricultural support tools, with potential integration of real-time environmental data, mobile platforms, and adaptive learning models to enhance decision-making in precision farming
Integrating K-Means Clustering and Apriori for Data Mining-Based Digital Marketing Strategy For Increasing UMKM: Study Case Stabat City
No full textMicro, Small, and Medium Enterprises (MSMEs) or UMKM in Bahasa are play a crucial role in regional economic development, yet they often face challenges in designing effective marketing strategies due to limited access to advanced analytical tools. Digital marketing supported by data mining offers a solution to this problem by enabling more precise customer segmentation and product bundling recommendations. This study aims to integrate K-Means clustering and Apriori association rule mining to develop data-driven marketing strategies for MSMEs in Stabat City, Indonesia, with a specific focus on rice sales data. A dataset consisting of 1,000 rice sales transactions was processed through a multi-stage methodology, including data preprocessing, clustering, and association rule generation. The Elbow and Silhouette methods suggested an optimal cluster number of k = 3, resulting in three distinct customer groups: (1) loyal high-value buyers, (2) price-sensitive buyers, and (3) premium-oriented buyers. Descriptive statistics highlighted differences in average transaction values, purchase frequency, and brand preferences across clusters. Apriori analysis produced the top ten significant association rules, such as {Medium Rice} → {Pandan Wangi Rice} with support = 0.14, confidence = 0.68, and lift = 1.23. Promotional simulations showed that generic discount campaigns could increase sales by approximately 3.0%, whereas targeted bundling strategies yielded smaller short-term gains (+1.53%) but offered stronger long-term potential, particularly for premium-oriented clusters. These findings are consistent with prior international studies, where customer segmentation combined with market basket analysis has proven effective for enhancing digital marketing outcomes. The study concludes that integrating clustering and association rules can provide MSMEs with actionable insights to optimize promotional strategies and improve competitiveness. However, limitations remain, including the relatively small dataset, reliance on manual parameter selection, and simplified modeling assumptions. Future research should expand to multi-sector datasets and explore advanced algorithms to validate and extend these findings
Application of LSTM-Based Deep Learning for Stock Return Prediction of DCII
No full textStock return prediction is one of the areas that has received great attention in modern finance because it can help investors make more informed decisions and reduce the risk of market uncertainty. This study applies a deep learning approach based on Long Short-Term Memory (LSTM) to predict the return of DCII (PT DCI Indonesia Tbk) shares as a representation of highly volatile stocks on the Indonesia Stock Exchange. The purpose of this study is to evaluate the performance of twelve LSTM variants—including LSTM-Base, LSTM-Wide, LSTM-Stack2, LSTM-Stack3, LSTM with Dropout, BiLSTM, BiLSTM with Attention, and LSTM with Attention Mechanism—by comparing their performance on daily (H=1) and weekly (H=7) prediction horizons using historical data from id.investing.com. The initial data undergo preprocessing involving local format cleaning, calculation of technical indicators (MA, EMA, MACD, RSI, ATR, Bollinger Bands, etc.), MinMax normalization, and sequencing (windowing) with 30, 60, and 120-day lookbacks. The training process uses a uniform configuration with Adam optimization and early stopping to prevent overfitting, while the evaluation employs MAE, RMSE, MAPE, and R² metrics. The results show that LSTM-Stack3 (LB=60, H=1) provides the best performance with MAE = 0.020, RMSE = 0.031, MAPE = 5.0%, and R² = 0.91, followed by LSTM-Stack2-DO as the second-best configuration. Meanwhile, the LSTM-LB120-H7—the only model evaluated with a seven-day horizon—achieves the lowest performance due to higher long-term uncertainty. These findings confirm that stacked LSTM architectures are more effective for short-term return forecasting, whereas longer horizons require hybrid or enhanced approaches for stable performance.
Comparative Performance Evaluation of MobileNetV3 and ResNet50 for Forest Fire Image Classification
No full textIndonesia is one of the countries with a high incidence of forest and land fires (karhutla), especially during the dry season, thus requiring a fast and efficient early detection system. This study aims to compare the performance of two popular deep learning architectures, namely MobileNetV3 (Large and Small variants) and ResNet50, in forest fire image classification tasks using a transfer learning-based approach. This study emphasizes the comparison between accuracy and computational efficiency in a CPU-only environment, which represents real-world conditions of use in the field without GPU support. The dataset used is a combination of local field images from the Puncak area, Bogor, and a curated public forest fire dataset to ensure the model's generalization ability to diverse geographical conditions. The results of the experiment show that ResNet50 provides the highest accuracy with a training accuracy value of 0.677 and a validation accuracy of 0.647, but requires longer training and inference times. Meanwhile, MobileNetV3-Large and MobileNetV3-Small showed better computational efficiency, with only slightly lower accuracy (0.635 and 0.61) and high training stability. These findings confirm that lightweight models such as MobileNetV3 strike an optimal balance between accuracy, speed, and resource consumption, making them an ideal solution for implementing edge computing-based early detection systems. Overall, this research contributes by providing an empirical comparative analysis that can serve as a reference for selecting deep learning architectures for efficient and adaptive forest fire detection systems that are constrained by hardware limitations
Integration of Machine Learning and Blockchain for Forest Fire Risk Prediction
No full textThis study presents an integrated framework combining machine learning and blockchain technology to enhance the accuracy, transparency, and reliability of forest fire risk prediction in tropical regions. Using geospatial and climatological datasets from Google Earth Engine (GEE), two ensemble algorithms—Random Forest (RF) and Extreme Gradient Boosting (XGBoost)—were trained to model spatial fire susceptibility based on variables such as temperature, humidity, rainfall, wind speed, and vegetation index (NDVI). The RF model effectively identified low-risk areas but was less sensitive to minority high-risk classes, while XGBoost demonstrated superior adaptability in handling class imbalance and achieved more balanced performance across all categories. To ensure data authenticity and traceability, the prediction results were validated and recorded on the Ethereum blockchain using smart contracts. Each prediction output was transformed into a cryptographic hash (SHA-256) to guarantee immutability and verifiability. The integration of machine learning with blockchain establishes a decentralized, tamper-proof, and verifiable prediction system, promoting data integrity and transparency in environmental monitoring. Overall, this research introduces a novel “verifiable prediction pipeline” that advances both artificial intelligence and blockchain applications in environmental informatics, supporting proactive and accountable forest fire mitigation strategies
MCDM-AHP and CODAS Collaboration Techniques for Selection of Expert Education Personnel
No full textEducational progress is largely determined by human resources who have the best qualifications, with the ability of human resources to provide hope for educational development for a creative and potential future. The ability of human resources to create various variants of knowledge that can be developed to enlighten the progress of thinking through education to create expert education personnel. The aim of this research is to provide techniques to guarantee the quality of the selection process for expert education personnel for the competitive progress of mastering educational technology who are able to independently increase the creative and potential thinking of graduates. To achieve this, of course, strict collaboration techniques are needed in the selection process to obtain expert education personnel. The method proposed in this research is MCDM-AHP in collaboration with CODAS. These two methods can collaborate in providing guarantees for an optimal selection process for education personnel through eight selected assessment criteria and twelve alternatives. From the results obtained, the highest priority was obtained by ALT10 with a weight of 0.229. This gain goes through the stages of normalizing criteria and alternatives with the optimization results of both. With the research results that have been described in detail, the collaboration of the MCDM-AHP and CODAS methods can be used as a measuring tool for optimal assessment of the acquisition of decision support results and can be used as a comparison with other methods for measuring the level of optimization of results.
 
Development of Mobile Application by Applying Content-Based Filtering
No full textThe rapid advancements in information technology have transformed modern lifestyles, driving changes in consumer behavior and expectations, especially in the retail industry. This study focuses on developing a mobile application for Ampu Mart, a newly established retail business in Indonesia, to optimize product recommendation systems using the Content-Based Filtering (CBF) approach. The research integrates CBF with string matching and cosine similarity algorithms to provide personalized product recommendations based on customer preferences, enhancing user satisfaction and supporting more efficient purchasing decisions. The methodology involves several stages, including problem identification through observation and interviews, data collection on product attributes and customer preferences, system design, prototype development, implementation, and testing. The application leverages advanced algorithms to analyze product characteristics, ensuring relevant recommendations by matching user preferences with product attributes. User Acceptance Testing (UAT) conducted with 30 participants—customers, administrators, and management—evaluated the application's functionality, usability, accuracy, and performance. Results indicate that the mobile application improves the shopping experience and boosts sales by offering accurate, user-centered recommendations. The findings highlight the strategic importance of integrating intelligent technology into e-commerce platforms to enhance competitiveness in the retail market. Future work recommends incorporating Collaborative Filtering techniques to further enrich the recommendation system by analyzing collective customer behavior
Enhancing Sentiment Analysis Accuracy Using SVM and Slang Word Normalization on YouTube Comments
No full textSentiment analysis is a crucial technique in understanding public opinion, particularly on social media platforms such as YouTube. However, the presence of informal language, including slang words, poses significant challenges to accurate sentiment classification. This study aims to enhance sentiment analysis by implementing a Support Vector Machine (SVM) classifier combined with SMOTEENN data balancing techniques to address class imbalance issues. The research collects 3,375 YouTube comments on the movie Pengabdi Setan 2: Communion using the YouTube Data API. The preprocessing steps include text cleaning, tokenization, stopwords removal, stemming, and slang word normalization using kamusalay.csv to ensure standardization of informal expressions. The extracted features are represented using TF-IDF, and sentiment labeling is performed using VADER. Experimental results show that the SVM model achieves an accuracy of 98%, but struggles with detecting negative sentiments, as indicated by lower recall (38%) and F1-score (53%) for the negative class. Although the application of SMOTEENN improves data balance, further refinements, such as adjusting classification thresholds and integrating deep learning techniques, are necessary to enhance sentiment detection, particularly for informal and emotionally nuanced language. This study contributes to improving sentiment analysis models by demonstrating the effectiveness of slang word normalization in handling non-standard language variations. Future work will explore more sophisticated language models to enhance accuracy in sentiment classification
Hybrid Deep Learning and USE Algorithm for Essay Scoring: Accuracy and Performance Analysis
No full textThe main challenge in digital education, particularly in the automatic assessment of essay answers in online learning systems, lies in the complexity of natural language understanding and semantic evaluation required to achieve the level of precision equivalent to human judgment. This study aims to develop and analyze the performance of a hybrid model that combines deep learning with a semantic similarity-based approach to essay auto-grading. The methods used include the collection of essay answer data from various disciplines, text processing to extract semantic representations, and the calculation of the degree of similarity between the participant's answers and the answer key using the similarity measure. The evaluation was carried out by comparing the results of automatic assessments with manual assessments by teachers. The results showed that the developed model achieved the highest accuracy level of 90.22% at 0.8 treshold, with a precision of 84.63%, a recall of 100%, and an F1 score of 91.68%. To strengthen the reliability of the findings, statistical validation was carried out using error evaluation metrics. RMSE value is 0.32 and RMAE value is 0.19. These findings show that the model is able to mimic human judgment reliably and consistently, and can distinguish linguistic variations that arise in different types of essay questions. This system offers an effective solution for the automation of assessments in an online learning environment, while maintaining the integrity and objectivity of the evaluation