Publication Search

The development of digital services cannot be separated from users' habits in using smartphones and the affordability of internet access. Smartphone users want services that can carry out various activities easily, including telecommunications. This opportunity was captured by telecommunications companies by creating provider applications for smartphones. In connection with this, PT. Telekomunikasi Indonesia Tbk (Telkomsel) launched a new provider product called by.U. The purpose of this research is to determine the factors that influence user satisfaction with the by.U application based on the End User Computing Satisfaction (EUCS) model from William J. Doll & Gholamreza Torkzadeh (1988) with six variables consisting of content, accuracy, format , ease of use, timeliness, and satisfaction. This research used simple random sampling with a total of 60 respondents who were residents of the city of Surabaya who used the by.U application. The research results show that the variables of content and timeliness have a direct effect on satisfaction. Meanwhile, the variables accuracy, format and ease of use have a positive but not significant effect on satisfaction. Timeliness is the factor that has the most influence on satisfaction seen from the results of hypothesis testing which produces the largest path coefficient value, which is 0.268.

Rapid technological developments encourage the banking sector to continue to innovate so as not to be left behind. Tight competition in this industry is caused by customers' freedom to choose products and services that are considered more profitable. This phenomenon is known as Customer Churn, which is a condition where customers choose not to continue subscribing to a particular company. The method applied uses a machine learning approach and customer segmentation approach. The churn analysis results show that the machine learning model, especially the random forest model, has the highest level of accuracy with an F1-Score of 91%. This model has the potential to reduce churn rates from 20.4% to 5.61%, illustrating its positive impact. Apart from that, for the clustering results, the K-Prototype model was obtained for the clustering model with the highest Silhouette Score number of 0.1557 and 4 clusters were obtained.    

The global economic recession is a global economic downturn that affects the domestic economies of countries in the world. The stronger the economic dependence of one country on the global economy, the faster a recession will occur in that country. In 2020 the country of Indonesia and even the world are exposed to the COVID-19 virus which has an impact on the country's economic growth, even the world economy. This is the trigger for an economic recession. This has led to many different public perspectives on the occurrence of a global economic recession whose opinions or reactions are expressed on social media Youtube. The data was obtained by crawling techniques from social media Youtube with a total of 500 comments used. The data is then labeled (class) with a lexicon-based method with an Indonesian language dictionary. From the labeling results, it was obtained 185 positive labeled data (37%) and 315 negative opinions (63%). The data preprocessing stage is carried out in preparation for the data to be processed for sentiment analysis. Of the many opinions obtained, an analysis of public sentiment regarding the 2023 global economic recession will be carried out using the Naïve Bayes classification algorithm. This study also applied the TF-IDF word weighting method with the n-gram feature used, namely bigram (n=1). The system will be evaluated using a confusion matrix. The implementation results show a prediction model with a total of 500 opinion data with a comparison of training data and test data of 9:1, producing an accuracy value of 84.00%, a precision value of 75.00%, a recall of 30.00%, and an f1-score of 42.86%. The performance of the system model built in this study can be said to be good.

Fraud detection is used in various industries, including banking institutes, finance, insurance, government agencies, etc. Recent increases in the number of fraud attempts make fraud detection crucial for safeguarding financial information that is confidential or personal. Many types of fraud problems exist, including card-not-present fraud, fake Marchant, counterfeit checks, stolen credit cards, and others. An ensemble feature selection technique based on Recursive feature elimination (RFE), Information gain (IG), and Chi-Squared (X2) in concurrence with the Random Forest algorithm, was proposed to give research findings and results on fraud detection and prevention. The objective was to choose the essential features for training the model. The Receiver Operating Characteristic (ROC) Score, Accuracy, F1 Score, and Precision are used to evaluate the model's performance. The findings show that the model can differentiate between fraudulent transactions and those that are not, with an ROC Score of 95.83% and an Accuracy of 99.6%. The F1 Score of 99.6%% and precision of 100% further sustain the model's ability to detect fraudulent transactions with the least false positives correctly. The ensemble feature selection technique reduced training time and did not compromise the model's performance, making it a valuable tool for businesses in preventing fraudulent transactions.

Indonesia is entering a critical period for mental health. Research results from the The Indonesia National Adolescent Mental Health Survey (2022), around 15,5 million Indonesian teenagers experience mental health disorders. Students are part of late adolescence and are vulnerable to mental disorders. The binary logistic regression model is used to examine in more depth what variables have a significant effect. So, this research aims to predict mental health of students in the Faculty of Health Sciences, Pekalongan University. This type of research is observasional with a cross-sectinal design. Data were collected using the SRQ-20 via the Google Form platform using simple random sampling of 186 students. There were 130 students who indicated mental health disordes (69,9%). Simultaneously age, gender, major, semester level, mother’s educational level, father’s educational level, social support and dependence on using smartphone influence student’s mental health status (P Value<0,05). Even though only a few variables were partially significant, the precision percentage of the model that could be predicted correctly was 71,5%. The accuracy of the predicted model is quite good, namely student mental health status (y) = -3,720 + 2,403 (Major) – 1,980 (Mother’s Educational Level) + 1,444 (Father’s Educational Level) + 0,888 (Dependence on using Smartphone). Promotive and preventive interventions such as further screening and education to support student’s healthy mental health.  

Batik culture is very strong in Indonesia, this is the reason that batik can be found throughout the archipelago, with unique characteristics that distinguish it in each region. However, people are often confused and find it difficult to recognize one type of batik from another. One of the famous types of batik motif is Batik Parang. This research aims to establish a Convolutional Neural Network (CNN) model to classify Batik Parang and help people distinguish it from other batik motifs. Deep learning, particularly CNN, was chosen because it has a high accuracy rate in image classification. A quantitative Experimental design is used, using a dataset of 100 batik images evenly divided into two classes, namely Batik Parang and not Batik Parang. The dataset is divided into two categories, namely training data and testing data, with a data ratio of 80:20. Thus, by using Convolutional Neural Network (CNN), the classification between Batik Parang and not Batik Parang produces an accuracy of 95%, with the use of epoch = 118 and batch_size = 100.

The Influence of Leadership Style, Organizational Commitment and Organizational Culture on PT SOS Indonesia Employee Job Satisfaction. Office Administration Education Undergraduate Study Program, Faculty of Economics, Jakarta State University.The aim of this research is to determine the influence of leadership style, organizational commitment and organizational culture on the job satisfaction of PT SOS Indonesia employees. The type of research carried out uses quantitative research methods. The data collection technique used was through a questionnaire which was measured using a Likert scale from a value between one to five with a total of 16 questions. The population in the study was 135 employees of PT SOS Indonesia with a sample size of 100 employees at PT SOS Indonesia. This research uses a non-probability sampling technique with purposive sampling with the Slovin formula with a difficulty level accuracy of 5%. The analysis technique used in this research is SEM PLS (Structural Equation Modeling - Partial Least Squares) which was carried out using SmartPLS 4.0 software. Data The analytical methods used in this research include measurement model analysis, structural model analysis, and direct influence hypothesis testing. The results of data analysis show that the leadership style variable has a positive and significant effect on job satisfaction, the organizational commitment variable has a positive and significant effect on job satisfaction and organizational culture has a positive and significant effect on job satisfaction.

The significant changes in daily life patterns, driven by technological advancements, particularly in the transportation sector, are evident through the emergence of on-demand services such as Gojek. This research aims to explore users' perspectives and opinions regarding service quality, focusing on aspects like driver behavior, responsiveness, and reliability within the Gojek platform. The Naive Bayes method is employed to analyze user sentiments toward the driver services, supported by the Orange software to comprehend the complex patterns in user reviews. Evaluation is conducted on reviews from the Play Store, resulting in an accuracy of 87.4%, F1 score of 87.6%, precision of 87.9%, and recall of 87.4%. These findings indicate the success of the model in identifying and predicting predefined variables. Through the combination of methods and software, the study concludes that sentiment analysis of Gojek's driver services can be performed efficiently and reliably, providing valuable insights for online motorcycle taxi service providers.

Malaria is a global health challenge with significant impact in various countries. This study aims to analyze the effect of data augmentation on the performance of Convolutional Neural Networks (CNN) in malaria classification. The dataset used involves red blood cell images, with the main focus on the detection of Plasmodium parasites as the cause of malaria. Two CNN models were implemented, namely the model without data augmentation and the model with data augmentation. The model without augmentation showed a test score of 94%, while the model with augmentation showed a significant performance improvement with a test score of 96%. The overall accuracy of the augmented model on the test set reached 94%, signifying a significant improvement in diagnostic accuracy.

Parkinson's disease (PD) is a neurodegenerative disorder causing a decline in dopamine levels, impacting the peripheral nervous system and motor functions. Current detection methods often identify PD at advanced stages. This study addresses early-stage detection using handwriting analysis, specifically exploring the PaHaW dataset for pen pressure and stroke movement data. Evaluating online and offline features, the research employs pre-trained CNN models (VGG 19 and AlexNet) for offline datasets, achieving an overall accuracy of 0.53. For online datasets, velocity, and acceleration features are extracted and classified using Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM), and recurrent neural networks (RNN), with GRU yielding the highest accuracy at 0.57. Notably, the convolution-based model C-Bi-GRU surpasses other architectures with a remarkable 0.75 accuracy, emphasizing its efficacy in early PD detection. These findings underscore the potential of handwriting analysis as a diagnostic tool for PD, contributing valuable insights for further research and development in medical diagnostics.

Brain tumors are abnormal cell growths in brain tissue that can be life-threatening. This study aims to classify brain tumors to help early diagnosis. The method used is to extract features from brain MRI images using Local Binary Pattern (LBP) and then classified with Support Vector Machine (SVM). The data used were 2044 brain MRI images consisting of 3 classes namely meningioma, no tumor, and pituitary. The best results were obtained using LBP with a radius of 1 and the number of neighbors 8, while the best SVM model used the RBF kernel with a C value of 50, resulting in 88% accuracy, 86% precision, and 87% recall. It can be concluded that the combination of LBP and SVM methods is effective enough to classify brain tumor types to support early diagnosis.

Umi Rahma Maternity Clinic which provides pregnancy and childbirth examination services. Eclampsia ranks second in causes of maternal death. Preeclampsia diagnosis, which is a precursor to eclampsia, needs to be implemented to reduce maternal and child mortality rates. The aim of this research is to create an expert system for early diagnosis of preeclampsia in pregnant women using the Forward Chaining method. Research methodology stages include literature study, observation, interviews, data collection, primary and secondary. data preprocessing, database design, system design, analysis the calculation results. ,Gestational Age, Maternal Age, Edema. From the system output results, namely the diagnosis of mild preeclampsia and severe preeclampsia. The use of this expert system is web-based using the PHP programming language. To find out the accuracy of the system, a system evaluation was carried out by comparing the expert results with the system results on 10 testing data, and getting 1 result indicating an incorrect result and 9 correct results. Thus, the system accuracy percentage is 90%. So this research was carried out by creating an expert system which aims to make it easier to access information regarding the symptoms of preerclampsia in pregnant women and is able to provide knowledge to pregnant women online to make it more accurate. This research uses the extreme programming method in developing an expert system which will be built using the PHP and MySQL programming languages ​​as the database and using Unified Modeling Langueage (UML) modeling.

The use of deep learning models has become a major focus in optimizing the efficiency of machine learning applications. This research discusses various deep learning models that can be applied to improve efficiency in the context of machine learning applications. These models are designed to handle the complexity of machine learning tasks with a high level of accuracy while still considering aspects of computational efficiency. This article involves an in-depth look at several deep learning models that have proven effective in various application domains. Discussion includes the use of convolutional neural network (CNNs) models for image processing, recurrent neural networks (RNNs) for sequential data, and transformer-based models for natural language processing tasks. In addition, deep learning model tuning and optimization strategies, such as pruning and quantization, are also discussed to improve the efficient use of computing resources. This research identifies challenges and opportunities in integrating these deep learning models into machine learning applications with maximum efficiency. By considering the need for accuracy and limited computational resources, this research provides a holistic view of the approaches that can be applied to deal with complexity in diverse machine learning scenarios. The results are expected to provide a significant contribution to the development of efficient and effective machine learning applications.

The use of the Decision Tree method in smartphone price classification is the focus of this study. By using the 10 most relevant features and data normalization to achieve scale consistency, the Decision Tree algorithm delivers an average accuracy of 81%. Although some false positives and false negatives occur, the model is able to classify smartphone prices well, especially in identifying low and high prices. These results provide important insights into the features that affect smartphone prices. While there is still room for improvement, this model provides a solid foundation for the smartphone industry to determine prices based on certain specifications. The importance of relevant feature selection and data normalization was revealed in this study. Despite the accuracy reaching 81%, improvements in the classification of medium and high price classes are still possible to reduce prediction errors. This method provides an important basis for the smartphone industry to set prices based on specifications, and data mining techniques such as Decision Tree can be improved to improve the accuracy of future price predictions.

This research aims to analyze and predict the level of air pollution in Jakarta City using ISPU data of DKI Province, adopting the Naïve Bayes method. The test results show that the Naïve Bayes algorithm has excellent performance, with 93% accuracy, 98% precision, 100% recall, and 99% f1-score. The implication is that this model can be effectively used for air pollution forecasting in Jakarta City, assisting authorities in making decisions related to air quality and environmental improvement efforts.

An audit is a systematic, independent examination of financial statements, accounting records, and supporting documents prepared by management, the purpose of which is to form an opinion on the accuracy of financial statements. Financial statements must have relevant characteristics (reliability) and reliability (reliability). Without the services of auditors, management cannot convince outsiders that the financial statements presented by management contain reliable and reliable information. Independence is the auditor's attitude to impartiality. The experience of the examiner contributes to high-quality inspection. The purpose of this study is to analyze the influence of auditor behavior, time pressure, audit experience, and independence on audit quality.   The research was conducted at a Public Accounting Firm (KAP) in Medan City. The number of research samples of 80 people was selected by the nonprobability sampling method. Data collection was carried out by questionnaire through Google form and literature studies that supported this study. The method used in this study is to use the Structural Equation Model (SEM) equation using the Partial Least Square (PLS) tool version 3.0. PLS consists of external relationships (outer model) and internal relationships (inner model), cross-loading> 0.7, Composite Reliability, Convergent Validity, Exploratory Factor Analysis (EFA), and Confirmatory Factor Analysis (CFA). Based on the results of the analysis, it was found that the first, second, fourth, and seventh hypotheses were rejected where each variable such as auditor behavior, and time pressure,  did not affect audit quality and independence could not moderate the influence between auditor behavior on audit quality,  independence could not moderate audit quality. The third, fifth, and sixth hypotheses are accepted where each variable such as audit experience, independence moderates the effect of time pressure on audit quality, and independence moderates the effect of audit experience on audit quality.

This community service activity shows that the Puskad business incubator as part of the community can remain closely connected with the community that gave birth to it. For example, the problem of poverty which continues to haunt social life must continue to be eradicated. This can be alleviated, for example, by entrepreneurship or business development. Meanwhile, business development continued to develop over time, resulting in the rise of social media. With this service, we can see that people must share knowledge and experience with each other so that we can develop together. Therefore, answers from academics to problems and society can be displayed here. Using the participatory action research method, it was found that the way to increase the younger generation's understanding and knowledge of social media is to start with interaction. The differences that arise from social media training can be seen through the accuracy of conveying information to the appropriate target audience, equipment and communication facilities between creative actors, and budget conditions to meet the training and creativity needs of the community. The results of content creator training for entrepreneurs are: 1) Improved Communication Skills; 2) Increased Brand Awareness; 3) Increase interaction with customers; 4) Increase Conversions and Sales; 5) Marketing Cost Savings

Butterflies’ recognition serves a crucial role as an environmental indicator and a key factor in plant pollination. The automation of this recognition process, facilitated by Convolutional Neural Networks (CNNs), can expedite this task. Several pre-trained CNN models, such as VGG, ResNet, and Inception, have been widely used for this purpose. However, the scope of previous research has been somewhat constrained, focusing only on a maximum of 15 classes. This study proposes to modify the CNN InceptionV3 model and combine it with three data augmentations to recognize up to 100 butterfly species. To curb overfitting, this study employs a series of data augmentation techniques. In parallel, we refine the InceptionV3 model by reducing the number of layers and integrating four new layers. The test results demonstrate that our proposed model achieves an impressive accuracy of 99.43% for 15 classes with only 10 epochs, exceeding prior models by approximately 5%. When extended to 100 classes, the model maintains a high accuracy rate of 98.49% with 50 epochs. The proposed model surpasses the performance of standard pre-trained models, including VGG16, ResNet50, and InceptionV3, illustrating its potential for broader application.

Abstract.Network anomaly detection is a situation that occurs in network traffic that causes conditions to become abnormal. This research aims to analyze the performance of various machine learning algorithms in network anomaly detection and compare the performance of single classifier algorithms with ensemble learning. This ensemble learning technique has advantages such as increased accuracy and performance, can reduce the risk of overfitting and underfitting by using different subsets and features of data, and can turn weak learning into strong learning. However, on the other hand, this ensemble learning technique also has disadvantages in its use, namely that this ensemble method may not work well with high variance models, as the ensemble method may not be optimized for anomaly detection and that this method can be computationally expensive and time consuming due to the need to train and store multiple models. Some of the techniques used are deep learning, eager learning, lazy learning, bagging, feature selection, boosting, and stacking. In addition to this, this machine learning algorithm has weaknesses, including if any of the data used is incomplete, it will result in inaccurate completion data, making the programming process quite time-consuming. This research can help develop a more effective and efficient network anomaly detection system. The results of this research show that using ensemble learning and feature selection techniques can improve anomaly detection performance by reducing the processing time of redundant data and classification, as well as increasing precision values.    

The purpose of this research is to determine the influence of learning motivation and style in learning on learning outcomes. The type of research that will be carried out is using quantitative research methods with descriptive analysis. Data collection was carried out through a questionnaire which was measured using a Likert scale from one to five for 100 respondents with the criteria being students aged 18-22 years. The data analysis technique that will be used in this research is the Outer Model with calculations of Convergent Reliability, Discriminant Validity, Composite Reliability, Cronbach's Alpha and Inner Model with T statistics, R-Square and VIF calculations using SmartPLS 4 tools. The results of the research show that learning motivation influences learning achievement because of the accuracy factor in collecting assignments, this has high motivation. Learning style has no effect on learning achievement because it is caused by the very minimal study time that students spend. Apart from that, learning motivation and learning style together influence student learning achievement.