Publication Search

The coastal area of Manado City, particularly the estuary of the Bailang River in Bunaken District, frequently experiences inundation due to the complex interaction between river discharge and tidal fluctuations. This study aims to analyze the spatial distribution and extent of flooding resulting from the combination of river discharge and tides, as well as to evaluate mitigation strategies based on numerical modeling results. A two-dimensional hydrodynamic model was developed using MIKE 21 Flow Model FM HD, utilizing topographic and bathymetric data from DEMNAS and BATNAS, along with 2023 river discharge and tidal data. Simulations were conducted for existing conditions, minimum discharge with maximum tide, and maximum discharge with maximum tide. The results show that tidal influence significantly contributes to the increase in flood extent and duration, especially when high river discharge coincides with peak tidal conditions. The most affected areas are Kelurahan Molas and Tumumpa, located along the riverbanks and near the fish auction port, with inundation covering up to 4.973 hectares and water depths exceeding 0.5 meters. Recommended mitigation strategies include river normalization and the implementation of a real-time hydrodynamic early warning system. This study supports flood risk management and coastal spatial planning, serving as a technical reference for local government policy-making.

Crime is a social issue that continues to evolve alongside increasing community activity and regional development. This study aims to Cluster crime data in Binjai City based on the location of incidents using the K-Means algorithm and the Cross Industry Standard Process for Data Mining (CRISP-DM) approach. The data were obtained from the Binjai Police Department, with attributes including the type of crime, time of occurrence, and location, categorized by district. A comprehensive data preprocessing stage was carried out, involving the extraction of information from raw data, normalization of crime type labels, and conversion of categorical data into numerical form using label encoding. The optimal number of Clusters was determined using the Silhouette score method, which yielded the best result at K = 10. The Clustering results were further evaluated using the Davies-Bouldin Index (DBI) to ensure Cluster quality. The analysis revealed that Binjai Utara District has the highest number of crimes, particularly aggravated theft (curat), which frequently occurs from early morning to late morning. This Clustering is expected to provide valuable insights for authorities in formulating more targeted and data-driven regional security strategies.

This research analyzes the ethics of broadcasting advertisements on television media, especially related to  broadcasting time regulations. The significant role of television in shaping people's thoughts and  behaviors, coupled with the proliferation of potentially unethical and inappropriate broadcasts, highlights  the urgency of this issue. Such violations, especially those concerning children and adolescents, pose  significant risks due to exposure to inappropriate content and the normalization of unethical behavior. This  study emphasizes the importance of media commitment to the Broadcasting Behavior Guidelines and  Broadcast Program Standards (P3SPS) and Law No. 32 of 2002, which aims to protect viewers from  harmful information. This study uses a library observation method with a qualitative approach, analyzing  existing reports, research, and written sources. Data analysis uses Miles and Huberman's qualitative  decomposition technique, which includes data reduction, data presentation, and drawing conclusions. The  findings of the study are in line with previous studies, indicating that violations of broadcasting ethics and  broadcasting hours are systemic problems that have not been resolved. This underlines the need for stricter  supervision, sanctions, re-evaluation of broadcasting time classifications, and media literacy education for  the community.

Attendance management is an essential component in educational institutions, companies, and organizations to monitor the presence and punctuality of participants. Traditional attendance systems, such as manual signatures or identification cards, are prone to various issues including human error, time inefficiency, and identity fraud. To address these challenges, this study aims to develop a smart attendance system using facial recognition technology based on Python and the OpenCV library. The system is designed to automatically detect and recognize faces in real time using a webcam or camera module. It employs computer vision techniques to capture facial images, extract unique features, and match them against a stored database of registered participants. Once the face is verified, the system records the attendance along with a timestamp, ensuring data accuracy and security. The development process involved several stages, including image acquisition, preprocessing, feature extraction, and classification. OpenCV was utilized for image processing tasks, while Python provided the programming framework to integrate all components. To enhance recognition accuracy, the system applied techniques such as histogram equalization for lighting normalization and Haar Cascade classifiers for initial face detection. An experimental evaluation was conducted under various conditions, including different lighting environments and facial orientations. The results demonstrated that the system achieved an accuracy rate of 96% under normal lighting conditions, with only a small decrease in performance under dim or uneven lighting. These findings indicate that the system is reliable for practical applications, especially in controlled environments. Conclusion: The Python-based facial recognition attendance system offers a more efficient, secure, and accurate alternative to conventional attendance methods. Future improvements may include the integration of deep learning models to enhance recognition robustness in diverse real-world scenarios.

Attendance management is an essential component in educational institutions, companies, and organizations to monitor the presence and punctuality of participants. Traditional attendance systems, such as manual signatures or identification cards, are prone to various issues including human error, time inefficiency, and identity fraud. To address these challenges, this study aims to develop a smart attendance system using facial recognition technology based on Python and the OpenCV library. The system is designed to automatically detect and recognize faces in real time using a webcam or camera module. It employs computer vision techniques to capture facial images, extract unique features, and match them against a stored database of registered participants. Once the face is verified, the system records the attendance along with a timestamp, ensuring data accuracy and security. The development process involved several stages, including image acquisition, preprocessing, feature extraction, and classification. OpenCV was utilized for image processing tasks, while Python provided the programming framework to integrate all components. To enhance recognition accuracy, the system applied techniques such as histogram equalization for lighting normalization and Haar Cascade classifiers for initial face detection. An experimental evaluation was conducted under various conditions, including different lighting environments and facial orientations. The results demonstrated that the system achieved an accuracy rate of 96% under normal lighting conditions, with only a small decrease in performance under dim or uneven lighting. These findings indicate that the system is reliable for practical applications, especially in controlled environments. Conclusion: The Python-based facial recognition attendance system offers a more efficient, secure, and accurate alternative to conventional attendance methods. Future improvements may include the integration of deep learning models to enhance recognition robustness in diverse real-world scenarios.

The Kanjuruhan tragedy on October 1, 2022, marks one of the deadliest disasters in Indonesian sports history, resulting in the deaths of over a hundred spectators during a post-match riot. This study explores the tragedy through the lens of sports sociology by examining how crowd behavior, institutional failures, and the use of excessive force reflect deeper socio-political dynamics. Using a library research method, this paper analyzes academic literature, official reports, and sociological theories—particularly those of Gramsci, Bourdieu, and Foucault—to understand the normalization of violence, the exercise of hegemonic power through sport, and the systemic weaknesses in match organization. The study finds that the convergence of commercial interests, inadequate crowd management, and institutionalized violence created the conditions for this disaster. The findings call for structural reforms in sports governance and advocate for a sociologically informed approach to fan education, crowd control, and ethical media representation.

Learners are continually faced with choosing appropriate courses or making career choices due to increased educational opportunities. The emergence of machine learning-based course and career recommender systems has the potential to address this issue, offering personalized course recommendations tailored to individual learning pathways, preferences, and learning history. The optimization and feature engineering techniques and practical deployment environments have not been collectively examined in the previous research, despite the significant advancements in this area of research. Furthermore, previous research has rarely synthesized how these technical components help students choose appropriate courses and careers. This systematic review was carried out to investigate the current state of machine learning-based course and career recommender systems, focusing on key elements, such as primary data sources, feature engineering methods, algorithms, optimization techniques, evaluation metrics, and the environments where the existing course recommendation models are deployed. The PRISMA method for conducting a systematic review was used to choose studies that met the requirements for inclusion and exclusion. The study findings show significant reliance on interpretable and traditional machine learning algorithms, such as K-Nearest Neighbor and Random Forest, to develop recommender models. Feature engineering remains basic, as most studies rely on normalization, while optimization processes are often underreported. Also, evaluation metrics varied widely, impeding comparability, while most of the recommender models are deployed in an e-learning environment, leaving the traditional learning environment underrepresented. Furthermore, the study findings identified issues including data sparsity and diversity, data security and privacy, and changes in learner preferences that may have an impact on the performance of recommender systems while recommending further studies to make use of standardized optimization methods, and automated domain-informed feature engineering frameworks, benchmark and annotated datasets in developing models the gives priority to learners’ success and educational relevance.

Handwritten digit recognition is one of the key challenges in the field of digital image processing and artificial intelligence, with significant potential in various applications such as automatic form input systems, handwritten data correction, and attendance systems based on handwriting. This study aims to develop a web-based information system capable of automatically recognizing handwritten digits using the K-Nearest Neighbors (KNN) classification method. The system is designed through several main stages, including image preprocessing (conversion to grayscale, thresholding, and image size normalization), feature extraction using the zoning technique, and classification using the KNN algorithm. This research utilizes the MNIST dataset, which contains thousands of handwritten digit images ranging from 0 to 9. The system is developed using the Google Colab platform, supported by Python libraries such as OpenCV, NumPy, and Scikit-learn. Test results show that the system can achieve an accuracy of over 90% at certain K values, indicating that the KNN method is quite effective and efficient in recognizing handwritten digit patterns. This system is expected to be applicable for various digitization needs of handwritten numbers in education, administration, and information technology sectors

This research analyzes user sentiment reviews of the KAI Access application from Google Play Store to improve customer service at PT Kereta Api Indonesia. The study uses a Natural Language Processing (NLP) approach with the Latent Dirichlet Allocation (LDA) algorithm to extract main topics from 10,000 reviews collected from April 2024 to April 2025. Analysis results show 40.7% positive sentiment reviews and 49.3% negative. After data preprocessing through case folding, normalization, tokenization, stopword removal, and stemming, seven optimum topics were found from negative sentiment with a coherence score of 0.508343 and two optimum topics from positive sentiment with a coherence score of 0.511673. Analysis based on five service quality dimensions (tangibles, reliability, responsiveness, assurance, and empathy) reveals that the reliability dimension becomes the main issue, including system instability, transaction failures, login difficulties, and data inaccuracy. The responsiveness dimension is the second priority, with users expecting fast and responsive service to complaints. The results of this study provide recommendations for PT KAI to prioritize improvements in system reliability and responsiveness aspects to enhance the overall user experience, which will ultimately impact customer satisfaction and loyalty.    

The online gaming industry continues to grow rapidly in Indonesia, with many users purchasing digital items through 3rd party top up services such as Pitopup.com. One of the main challenges faced by Pitopup.com is the difficulty in classifying the sales of each available game item. This research aims to apply the K-Nearest Neighbor (KNN) method to predict the sales classification of game items in order to find out the sales category for each game item and hopefully help increase stock efficiency. The dataset used was obtained from historical sales data on Pitopup.com from June to September 2024. The research stages include data processing, normalization using Min-Max Scaling, data transformation using label encoding, separating test and training data using a ratio of 80:20, and using confusion matrix as a model evaluation. The test results show that KNN algorithm is able to classify game item sales on the Pitopup.com website with a level of accuracy in several categories: marketable category at 100%, the moderately sellable category at 100% and the not sellable category at 100%.

Aspect-based sentiment Analysis (ABSA) is vital in capturing customer opinions on specific e-commerce products and service attributes. This study proposes a hybrid deep learning model integrating Bi-Directional Gated Recurrent Units (BiGRU) and Bi-Directional Attention Flow (BiDAF) to perform aspect-level sentiment classification. BiGRU captures sequential dependencies, while BiDAF enhances attention by focusing on sentiment-relevant segments. The model is trained on an Amazon review dataset with preprocessing steps, including emoji handling, slang normalization, and lemmatization. It achieves a peak training accuracy of 99.78% at epoch 138 with early stopping. The model delivers a strong performance on the Amazon test set across four key aspects: price, quality, service, and delivery, with F1 scores ranging from 0.90 to 0.92. The model was also evaluated on the SemEval 2014 ABSA dataset to assess generalizability. Results on the restaurant domain achieved an F1-score of 88.78% and 83.66% on the laptop domain, outperforming several state-of-the-art baselines. These findings confirm the effectiveness of the BiGRU-BiDAF architecture in modeling aspect-specific sentiment across diverse domains.

This study aims to classify chicken eggs based on their physical size using the concept of computer vision and image segmentation techniques. Compared to the standard methods that have been used so far, this alternative technology is expected to help standardize measurements, cost efficiency, and work effectiveness. In this study, the classification of chicken eggs was carried out using image segmentation and regression analysis. Thus, it is expected that the classification of chicken eggs will have increasingly accurate values. After the image is taken using a webcam, the image segmentation process is used to divide the image into homogeneous areas based on the RGB (true color) color intensity similarity standard. Regression analysis is used to study and measure the relationship between the number of pixels and the weight of the object. The number of pixels indicating the area of ​​the object is the result of image segmentation, which will be entered into the regression equation to calculate the weight (grams). The results showed that the color characteristics of chicken eggs have a normalization of R at least 0.41 and a normalization of G at least 0.3. In addition, the classification test has an accuracy of 100% (36/36) and a weight estimation accuracy of 42 percent (15/36).

The rapid development of information and communication technology allows society to access various information needed in daily life. The Law of the Republic of Indonesia Number 23 of 2006 concerning population administration serves as an important element in population management. Population documents are issued by official institutions and have legal legitimacy as valid evidence. The method used in this research regarding public sentiment towards e-ID card services is the survey method, which aims to collect data from a large population using a smaller sample. The steps or processes in this research using the SVM method consist of case folding, cleaning, tokenizing, normalization, stopword removal, and stemming. Based on the classification of 150 test data using SVM, the number of positive sentiments recorded is 110 opinions, while negative sentiments recorded are 40 opinions.

Workload analysis provides an opportunity to smooth workload or reduce work activities that do not provide added value. Detailed evaluation and analysis of each work element of a position will display a more detailed workload map. PT. XYZ requires detailed workload mapping to be able to display the load profile for each position. The results of this load mapping can be used for workload normalization analysis. This research uses the Full Time Equivalent (FTE) method. The data used in this research are job descriptions, frequency and working hours. The research results show that the FTE-based workload value of the Vacum Geiss AG operator is FIT (value 1-1.28), namely on 03-10-2023 it is 1.030 and on 04-10-2023 it is 1.030 which is fit because the FTE value is between The value is 1-1.28 so that the operator's work load is appropriate on that day, then on 05-10-2023 on the Geiss AG vacuum machine it is 0.915, it is still underload because the FTE value is below the value of 0.99 so the operator on that day is underloaded. the work is not suitable and additional work is needed.

This study presents the implementation of Transfer learning using the ResNet-18 architecture for classifying 10 musical instrument categories based on visual representations of audio signals. The audio waveform is transformed into image-like inputs appropriate for CNN processing, accompanied by data augmentation and ImageNet-standard normalization. ResNet-18 is utilized due to its efficient feature extraction capability enabled by residual blocks, which help overcome vanishing gradient issues. The model was trained for 10 Epochs using the AdamW optimizer and Cross-Entropy Loss. Experimental results show that the model achieved a maximum validation accuracy of 77.35%, with a stable downward trend in training loss, indicating effective feature learning. However, several misclassification cases were observed, particularly among instruments with similar spectral characteristics, such as drum–violin and tabla–sitar. These findings demonstrate that while ResNet-18 performs reliably for musical instrument classification, further improvements remain possible through deeper architectures like ResNet-50, more comprehensive hyperparameter optimization, and the use of richer audio representations such as Mel-Spectrograms. This research provides an essential foundation for developing automated music analysis systems powered by Deep Learning.

Diabetes Mellitus (DM) is a metabolic disorder characterized by high blood sugar levels due to insulin deficiency. Factors causing Diabetes Mellitus (DM) are lifestyle which includes diet, lack of exercise, monitoring blood sugar, and medication. Most people do not realize that they have DM and only find out when they experience severe symptoms. To avoid this, the k-Nearest Neighbor (KNN) method can be used to predict the possibility of developing diabetes. The aim of this research is to classify diabetes mellitus using the K-Nearest Neighbor (KNN) method and make people more aware of the risk of disease through healthy lifestyle changes. Data received from the Dharma Husada Clinic is categorized based on researchers' needs, including age, BMI, insulin, skin thickness, glucose, diabetes, genetics, and insulin. This research was carried out in three main steps: dataset input, preprocessing, and evaluation. The first stage is data analysis which begins by entering a dataset to train and test the model, where each data element has certain characteristics (attributes) and classes. Preprocessing steps include training data generation and data cleaning, which includes sanitization, lowercase, normalization, stopwords, stemming, and tokenizing. The final step is evaluating. Evaluation includes building an evaluation model and measuring the level of accuracy, building a predictive model, and saving the model. This research shows that the K-Nearest Neighbor (KNN) method can be used to classify diabetes mellitus (DM), but especially in a small dataset consisting of 245 dates and 8 attributes it is not accurate for patients aged 30 years. . A k value that is too small can cause overfitting, and a k value that is too large can cause underfitting. However, if the amount of data is small, the choice of k can have a large impact.    

This qualitative study investigates the impact of streamers’ swearing on the morality of Vietnamese teenagers, a topic of increasing importance in the era of digital streaming and online entertainment. The research was conducted through semi-structured interviews with nine Vietnamese teenagers, providing in-depth insights into their perceptions and attitudes towards the language used by streamers and its influence on their behavior. The thematic analysis of the interview data revealed four key themes: normalization of swearing in digital spaces, delineation between online and offline behaviors, influence of streamers as role models, and critical reflection and selective adoption. The study found that while swearing by streamers was normalized in digital contexts, participants also demonstrated a clear distinction between acceptable behaviors in online and offline environments. Streamers were often viewed as influential role models, affecting the language and attitudes of the adolescents. However, participants also engaged in critical reflection, selectively adopting behaviors that aligned with their personal and cultural values. These findings contribute to the understanding of digital media’s impact on adolescent development, particularly in non-Western contexts. The study underscores the complex interplay between cultural norms, digital media consumption, and moral development, highlighting the need for comprehensive digital literacy and ethical content creation in the digital age.

This research aims to predict customer churn in a telecommunications company using Logistic Regression (LR) and Gradient Boosting Classifier (GBC) algorithms. Customer churn poses a significant challenge as acquiring new customers is costlier than retaining existing ones. The dataset from Kaggle comprises 7043 records and 21 attributes. The process includes data pre-processing, cleaning, transformation, and normalization using a Min-Max Scaler. The data is split into features (X) and target (y), then divided into training and testing sets with an 80:20 ratio. Both models were trained and evaluated using a confusion matrix. Results show that the GBC model outperforms the LR model, with an accuracy of 83% compared to LR's 81%. This study demonstrates the effectiveness of GBC in predicting customer churn.

This study aims to apply the K-Nearest Neighbors (KNN) algorithm to predict stunting status in young children based on height and weight data. Stunting is a growth failure condition caused by chronic malnutrition that negatively impacts children's physical and mental development. The dataset includes height, weight, and stunting status of children. The results show that the KNN model with k=3 achieved 100% accuracy on the test data. Evaluation using the confusion matrix and classification report indicates perfect precision, recall, and F1-score for each class. Data normalization with StandardScaler improved the model's performance by ensuring all features are on the same scale. The KNN algorithm proves to be a simple yet effective method for predicting stunting, demonstrating significant potential for early detection and health intervention in children. This study recommends using a larger and more diverse dataset, as well as incorporating additional relevant features to enhance model accuracy. Implementing the model in a web or mobile application is also suggested to assist healthcare professionals in the field.

A substation is a transmission installation that distributes power to a load in a certain area. The Sekayu 150 KV Substation is the center for controlling electrical power load requirements and functions as a center for securing electrical power system equipment and as a center for the normalization process for disturbances in the Sekayu transmission area. The electrical power distribution system does not rule out the possibility of disturbances, especially disturbances caused by nature. If there is an unpredictable disturbance, appropriate and reliable safety equipment (protection system) is needed to be able to increase the distribution of electrical power to the load (consumer).