Sinkron : jurnal dan penelitian teknik informatika
Not a member yet
1261 research outputs found
Sort by
Implementation of the K-Means Method for Clustering Regency/City in North Sumatra based on Poverty Indicators
No full textPoverty has many negative effects on people's lives, such as difficulty meeting basic needs, limited access to adequate health and education services, and limited economic opportunities. North Sumatra faces significant poverty problems as one of the largest provinces in Indonesia. This requires special attention and a thorough investigation. Reducing poverty is a very important issue for the government of North Sumatra Province. Poverty-alleviation strategies can no longer be applied uniformly. Instead, it is necessary to consider all the factors that cause poverty in each region. This means that the approach that must be given to each regency or city based on its poverty level must be adjusted. To overcome this problem, clustering must be carried out to identify areas with different levels of welfare. The aim of this research is to cluster regencies and cities in North Sumatra Province using the K-means method based on poverty indicator variables. This research only uses three poverty indicators: gross regional domestic product, human development index, and unemployment rate. The optimal number of clusters is determined based on the results of the silhouette coefficient. The research method begins with dataset collection, exploratory data analysis, data preprocessing, and k-means clustering. The value k = 6 produces a silhouette coefficient of 0.4135. This research produced six regency/city clusters. Cluster 1 consists of 11 regencies and 1 city; cluster 2 consists of 1 regency and 2 cities; cluster 3 consists of 4 regencies; cluster 4 consists of 7 regencies; cluster 5 consists of 4 cities; and cluster 6 consists of 2 regencies and 1 city. The variables gross regional domestic product, human development index, and unemployment rate have a big influence on the cluster results. This will enable the government to adopt policies to tackle poverty quickly and effectively
Analysis of Service Satisfaction Level at Batam Center International Port in Batam City Using the Fuzzy Mamdani Method
No full textA port is defined as a facility and system where a port is a work environment consisting of a land and water area equipped with berthing and mooring facilities for ships, a place for loading and unloading, and boarding and disembarking passengers from one mode of sea transportation to another mode of transportation. In carrying out activities at the port, apart from providing sufficient facilities and equipment, quality services cannot be neglected. This research aims to analyze and determine how much influence the level of service provided has on the satisfaction response of ship passengers at the Batam Center International Port, Batam City. To carry out this research, the study used a literature review, observation, and distribution of questionnaires using a Likert scale and the Fuzzy Mamdani method. However, several challenges were encountered during the study. First, there was a difficulty in obtaining a representative sample of passengers due to the variability in passenger traffic and schedules. Second, the subjective nature of satisfaction measurement posed challenges in ensuring the accuracy and consistency of responses. Third, integrating the qualitative data from the Likert scale with the quantitative Fuzzy Mamdani method required careful handling to avoid misinterpretation The expected results from this research are the levels of satisfaction ship passengers feel regarding the dimensions of tangibles, reliability, responsiveness, assurance, and empathy. This will help determine whether passengers perceive themselves as satisfied or dissatisfied. Based on the analysis and discussion carried out with 50 respondents using the Likert scale method, the respondents stated that they were quite satisfied. This indicates that the Batam Center International Port, Batam City, is quite successful in providing satisfactory services to passengers. From the dimensions of reliability, responsiveness, assurance, empathy, and tangibles, overall consumers feel quite satisfied, with an average satisfaction percentage of 71% (35.5 out of 50 respondents). Similarly, the results obtained from the Fuzzy Logic Mamdani method show an output value for consumer satisfaction of 150 (within a range of 50–250), indicating that the level of consumer satisfaction with the services provided by the Batam Center International Port, Batam City, is quite satisfactor
Application of the XGBoost Model with Hyperparameter Tuning for Industry Classification for Job Applicants
No full textThe development of technology and changes in job market dynamics have created new challenges in aligning education with industry needs. In this research, the XGBoost model with hyperparameter tuning was applied for industry classification on job applicant data taken from the Kaggle dataset LinkedIn Job Postings in 2023. This dataset consists of 23 attributes with a total of 33,085 job vacancy data points. The experimental results show that both the model without hyperparameter tuning and with GridSearchCV produce the same classification accuracy, which is 0.89 or 89%, with stable precision, recall, and F1-Score values. The best parameters found in this study are colsample_bytree = 1.0, learning_rate = 0.3, max_depth = 6, min_child_weight = 1, n_estimators = 100, and subsample = 1.0. However, cross-validation using k-fold shows a significant increase in accuracy to 0.90, or 90%. This finding confirms that the use of cross-validation can improve the performance estimation of the model more accurately and robustly by utilizing all available data for training and testing. Moreover, the implementation of cross-validation demonstrates the importance of leveraging all data points to enhance model reliability and robustness. Future research can explore alternative hyperparameter tuning methods and apply the model to larger datasets to further validate the generalizability and reliability of the XGBoost model in different application contexts. Thus, this study underscores the significance of rigorous model evaluation techniques in achieving high-performing machine learning model
Edge Computing Architecture Sensor-based Flood Monitoring System: Design and Implementation
No full textThe purpose of this research is to develop and execute a system for monitoring floods using sensors and edge computing architecture. The goal is to make flood detection and prediction more accurate and faster. The growing frequency and severity of flood disasters in different parts of the world has prompted the necessity for a better system to track these events. The primary goal of this study is to design a system that can reduce network load and latency by processing sensor data locally at edge devices before sending it to the cloud. To detect and anticipate flood events, the research method incorporates several environmental sensors that measure things like soil moisture, water level, and rainfall. These readings are subsequently processed by edge nodes using machine learning algorithms. Compared to more conventional methods that depend only on cloud computing, the results demonstrate that the system can deliver quicker and more accurate predictions. Results showed a detection and prediction accuracy of 98.95% for floods. Edge computing also succeeded in drastically cutting down on bandwidth consumption and communication latency. This research concludes that edge computing architecture based on sensors can effectively monitor floods and has excellent potential for use in many different areas prone to flooding. Improving the prediction algorithm and investigating its potential integration with a more thorough early warning system should be the focus of future research
A Comparative Study of Alternative Automatic Labeling Using AI Assistant
No full textThe development of AI assistants has become increasingly sophisticated, as evidenced by their growing adoption in assisting humans with various tasks. In particular, AI assistants have demonstrated potential in the field of sentiment analysis, where they can automate the labeling of text data. Traditionally, this labeling process has been performed manually by humans or using tools like the VADER Lexicon. This study is imperative to evaluate the performance of AI Assistants in sentiment labeling, as compared to traditional human-based labeling and the application of the VADER sentiment analysis algorithm. The methodology involves comparing the labeling results of Gemini and You AI with those of human labeling and VADER. Performance is evaluated using the Naive Bayes and K- Nearest Neighbour algorithms, and K-Fold Cross Validation is employed for evaluation. The results indicate that the performance of both AI assistants can closely approximate the performance of human labeling. Gemini's best accuracy is achieved with the k-NN algorithm at 53.71%, while You AI's best accuracy is achieved with the Naive Bayes algorithm at 48.30%. These results are close to the accuracy of human labeling (61.12%) using the k-NN algorithm and VADER (54.29%) using the Naive Bayes algorithm. This suggests that AI assistants can serve as an alternative for text data labeling, as the differences in performance are not statistically significant
Application of the Arima Method to Prediction Maximum Rainfall at Central Java Climatological Station
No full textThe existence of extreme weather that is difficult to predict results in frequent hydrometeorological disasters. ARIMA is a prediction method that can capture trend patterns, seasonal cycles, and random fluctuations that are often found in patterned data. Although many samples of rain data collection points are needed to produce denser data, one point can be considered to represent an area that is not too large, such as Semarang City. This method is quite accurate for short-term forecasts, with the results of monthly maximum rainfall forecasts in 2023 showing varying MAPE values. For the 12-month forecast, prediction results range from fair to very accurate. The 7-month forecast also shows decent to very accurate results. However, the 5-month forecast shows less accurate results. This shows that ARIMA can be a useful method in forecasting monthly maximum rainfall, especially during the dry season. The application of ARIMA in Semarang City can help in planning hydrometeorological disaster mitigation, considering that the Semarang City area often experiences extreme weather that is difficult to predict. Thus, the use of ARIMA can provide significant benefits in preparing for and reducing the impact of hydrometeorological disasters in the region. In addition, with more accurate forecasts, the government and society can take preventative steps earlier, such as better water management, creating an adequate drainage system, and increasing public awareness of the threat of disasters. Therefore, this research emphasizes the importance of using reliable prediction methods such as ARIMA to improve preparedness in dealing with hydrometeorological disasters
Clustering Analysis of Socio-Economic Districts/Cities In East Java Province Using PCA And Hierarchical Clustering Methods
No full textThis study aims to analyze the socio-economic conditions of districts/cities in East Java using Principal Component Analysis (PCA) and Hierarchical Clustering. Socio-economic data for 2023 from 38 districts/cities includes the percentage of poor people, regional GDP, life expectancy, average years of schooling, per capita expenditure, and unemployment rate. PCA was used to reduce the dimensionality of the data, facilitating analysis and visualization. The reduced data was then analyzed using Hierarchical Clustering to group districts based on similar socio-economic characteristics. The clustering results were evaluated with the Silhouette Index and Davies-Bouldin Index. This study identified four main clusters with different socio-economic characteristics. The best clusters have high regional GDP, life expectancy, average years of schooling, and high per capita expenditure and low unemployment rates. The worst clusters show a high percentage of poor people and high unemployment rates. These results assist the government in designing more effective policies to improve welfare in East Java
Inorganic Waste Detection Application Using Smart Computing Technology with YOLOv8 Method
No full textWaste and renewable energy are critical issues in Indonesia, with the government aiming for renewable energy (RE) to contribute 23% to the national energy mix by 2025. This research focuses on developing a waste processing system through the TOSS (Tempat Olah Sampah Setempat) method and Peuyeumisasi technique to convert waste into biomass, such as briquettes and pellets for fuel. However, manual waste sorting remains time-consuming, prompting the need for a real-time detection system. You Only Look Once (YOLO) is an object detection approach that utilizes Convolutional Neural Networks (CNN) for object detection, making it one of the applications of intelligent computing in the field of computer vision. the latest version of YOLO is YOLO v8 offering several improvements over the previous version, can be employed in a real-time detection system to separate organic and inorganic waste. In this study, the dataset used consists of 2.000 images comprising five classes of inorganic waste: plastic bottles, plastic, glass, cans, and Styrofoam. The study demonstrates that YOLOv8 performs exceptionally well in detecting inorganic waste, with an average accuracy of 98% based on direct testing, and model evaluation showing an average accuracy of 99.33%, precision of 99.63%, recall of 96.53%, and an f1-score of 98.03%. These results indicate that the YOLOv8 method can significantly accelerate and simplify the waste sorting process, thereby supporting the conversion of waste into renewable energy. This research is expected to provide a practical solution and serve as a reference for future studies.
 
A Mixed-Integer Programming Approach on Clustering Problems with Segmentation Application Customer
No full textAs a marketing strategy, segmentation involves categorizing customers into specific groups based on their loyalty to a brand. This process is crucial in shaping an effective business strategy, as identifying various customer types enables businesses to target their marketing efforts more precisely. This research focuses on solving the cluster optimization problem by applying a combinatorial optimization approach to develop a cluster optimization method. The combinatorial optimization utilized here operates on a binary system, using 0s and 1s to identify the optimal cluster for each object. Specifically, a value of 1 indicates that an object is assigned to an optimal cluster, while a value of 0 signifies that the object belongs to a non-optimal cluster. By designating clusters with a value of 1, the method ensures that the best optimization value is achieved. The 0-1 non-linear problem model ensures that objects with the shortest distances between them are grouped in the same cluster. Additionally, the model guarantees that each object belongs to only one cluster and that, across k tests, every cluster contains at least one object. This model can also be used to determine the ideal number of clusters for a given dataset, ensuring optimal segmentation results for business applications
Implementation of Deep Learning Model for Classification of Household Trash Image
No full textThe problem of household waste management is a very important issue today, where the rapid urbanization, consumptive culture, and the tendency to dispose of waste without sorting it first from home, makes the volume of waste in landfills increase. Therefore, household waste management needs to be managed quickly and appropriately, so as not to have a major impact on environmental, hygiene, and health problems. Although some environmental communities and local governments have made efforts to manage waste through recycling systems, the long-term use of human labor is inefficient, expensive, and harmful to workers' health. Therefore, utilizing artificial intelligence technology is the best solution to classify waste types quickly and accurately. This research tries to test several pre-trained convolutional neural network (CNN) models to perform classification. The results of testing pre-trained CNN models, such as AlexNet, VGG16, VGG19, ResNet50, and ResNeXt50, found that the pre-trained model ResNext50 is better with 100% accuracy, while the training loss and validation loss are 0.0414 and 0.0304, respectively. Then the second best model is the pre-trained ResNet50 model with 100% accuracy with training loss and validation loss of 0.0832 and 0.1077, respectively