ejournal.nusamandiri.ac.id (STMIK Nusa Mandiri)
Not a member yet
1504 research outputs found
Sort by
COMPARISON OF DEEP LEARNING METHODS ON SENTIMENT ANALYSIS USING WORD EMBEDDING
Get PDFAccording to ICW, corruption cases in Indonesia in the last 5 years have increased and the amount of losses suffered by the state from 2012-2022 reached Rp138.39 trillion. According to Transparency International, Indonesia's CPI ranking decreased in 2023 to 115 compared to 2022 at 110 out of 180 countries. These results show that the response to corruption is still slow and continues to deteriorate due to a lack of support from stakeholders. The purpose of this study is to test and compare the performance of deep learning model algorithms (RNN/LSTM/GRU/Bi-GRU/Bi-LSTM) on sentiment classification using word embedding, and obtain a model architecture that can determine the polarity of a sentence about public sentiment related to corruption in Indonesia, which can help governments, researchers, and practitioners in designing more effective anti-corruption strategies. The dataset used amounted to 1793 derived from crawling Twitter with 3 classes namely positive, negative and neutral. This research starts from data collection, preprocessing, word embedding, splitting the dataset which is divided into 80% training data and 20% test data, deep learning model testing, model evaluation and result representation. Word embedding uses word2vec with a dimension of 300. Based on the experimental results obtained, Bi-GRU has better performance than other architectural models with an accuracy value of 88%, precision 88.07%, recall 86.97% and f1-score 87.51%. The data used in this research is relatively small, it is recommended that future research can overcome i
PERBANDINGAN PENERAPAN ALGORITMA DEEP LEARNING DALAM PREDIKSI HARGA EMAS
Get PDFDigital investment is trending because advancements in information technology make access easy through smartphones. Various digital investment instruments attract much interest from the public. Post COVID-19 pandemic, the economic impact of the pandemic is still felt until the end of 2022, requiring people to be smart in managing their finances. Gold investment is considered profitable due to its high value and tendency to increase, unlike the fluctuating stocks. Although easily accessible, investments carry risks, so investors must have sufficient knowledge to maximize profits. This research aims to predict gold prices using several deep learning models, namely Artificial Neural Network (ANN), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), and Long Short-Term Memory (LSTM). The dataset used was taken from the Kaggle website, which includes historical gold price data. In this research, various deep learning models were applied and evaluated to determine the best model for predicting gold prices. The results show that the CNN model with Adam optimization and Mean Squared Error (MSE) loss function provides the best performance. The CNN model achieved the lowest Mean Absolute Error (MAE) of 0.004848717761305338, the lowest MSE of 4.3451079619612133, and the lowest Root Mean Squared Error (RMSE) of 0.006591743291392053. These results indicate that the CNN model is more effective in predicting gold prices compared to the ANN, RNN, and LSTM models on the used dataset.
 
OPTIMIZATION OF POTATO LEAF DISEASE IDENTIFICATION WITH TRANSFER LEARNING APPROACH USING MOBILENETV1 ARCHITECTURE
Get PDFDiseases affecting potato leaves frequently lead to significant setbacks for farmers, reducing the overall harvest and the quality of the potatoes. Given the critical need for prompt disease detection, this research introduces the use of the MobileNet framework grounded in the Convolutional Neural Network (CNN) for adept detection of potato leaf ailments. During the research, potato leaf images undergo processing, and their distinct features are gleaned using CNN. Then, harnessing the MobileNet framework, these images undergo classification to ascertain the existence of diseases. The aspiration is that the formulated model can pinpoint diseases with notable precision, rapid feedback, and enhanced computational adeptness. Initial findings underscore the potential of this methodology in discerning potato leaf diseases, providing renewed optimism for farmers grappling with plant health issues. Experiments using the Transfer Learning approach showed good performance in classification and displayed a high accuracy rate of 99.2%
CLASSIFICATION OF CUSTOMERS’ REPEAT ORDER PROBABILITY USING DECISION TREE, NAÏVE BAYES AND RANDOM FOREST
Get PDFLimited customer information in sales data on e-commerce in Indonesia hinders companies in determining targeted marketing strategies, especially in targeting groups of potential customers to make repeat purchases. Sales data in the form of customers' names and cellphone numbers has been hidden by e-commerce, and only data is available in the form of products purchased, number of purchases, and customer addresses. So far, the methods used to determine potential customers mostly use more complete data features. Research that uses limited e-commerce data to determine potential customers is scarce. Several algorithms for predicting repeat purchases in e-commerce also have been widely used. However, the comparison of the performance of these methods in the context of e-commerce in Indonesia with limited data has yet to be discovered. In this research, the Decision Tree, Naive Bayes, and Random Forest methods were compared to classify potential customers using Maschere brand sales data from two e-commerce sites, namely Tokopedia and Shopee. The research results show that the Decision Tree algorithm achieved an accuracy of 90.91%, Naive Bayes achieved an accuracy of 37.50%, and Random Forest achieved the best level of accuracy, namely 93.94%. These results show that the Random Forest method is the best method for classifying customers' probability of repeat purchases. In the future, the results of this research can be developed again as a decision-making system to determine potential customers
UNDERSTANDING THE CONTINUANCE OF ELECTRONIC PAYMENTS USAGE AFTER COVID-19: A SURVEY IN INDONESIA
Get PDFDuring the ongoing pandemic with elevated COVID-19 cases, efforts to minimize direct physical contact for virus prevention have been heightened. Consequently, there has been a strong emphasis on adopting non-cash transactions, particularly electronic payments. As the Indonesian government revoked the social restriction policy on December 30, 2022, people gradually resumed normal activities such as work, school, and shopping. The question arises whether the widespread adoption of electronic payments will persist after COVID-19. To understand this and the factors influencing the sustained use of electronic payments, this study utilized the UTAUT, Trust, and Perceived Security as the research model. The findings indicate that all 920 survey participants maintain their electronic payment usage after COVID-19. Through PLS-SEM analysis, key factors contributing to the sustained use of electronic payment after COVID-19 include the intention to use electronic payments, user trust, performance expectations, facilitating conditions, and perceived security. Additional variables proposed in this research, user trust and perceived security, are proven to have an influence on users' intentions to continue using electronic payments
A COMPARATIVE EVALUATING NUMERICAL MEASURE VARIATIONS IN K-MEDOIDS CLUSTERING FOR EFFECTIVE DATA GROUPING
No full textThe K-Medoids Clustering algorithm is a frequently employed technique among researchers for data categorization. The primary difficulty addressed in this investigation pertains to the extent of optimality achieved when varying distance computation methodologies are applied within the framework of K-Medoids Clustering. This study is primarily concerned with the application of K-Medoids Clustering, employing a multitude of distance calculation methods, specifically those involving numerical metrics. The aim is to undertake a comparative analysis of Davies-Bouldin Index (DBI) values in order to ascertain the most productive distance calculation technique. In this research, the distance calculation methodologies include Manhattan Distance, Jaccard Similarity, Dynamic Time Warping Distance, Cosine Similarity, Chebyshev Distance, Canberra Distance and Euclidean Distance. The dataset consists of sales data from Devi Cosmetics, covering the period between January and April 2022 and comprising 56 distinct sales items. The research provides an exhaustive evaluation of numerical metrics concerning the K-Medoids Clustering algorithm. The findings indicate that the optimal clustering is achieved using the Chebyshev distance, resulting in 9 clusters with a DBI value of 166.632. The study's contribution is that it can improve more optimal data grouping to help make decisions correctly
UTILIZING RETRIEVAL-AUGMENTED GENERATION IN LARGE LANGUAGE MODELS TO ENHANCE INDONESIAN LANGUAGE NLP
Get PDFThe improvement of Large Language Models (LLM) such as ChatGPT through Retrieval-Augmented Generation (RAG) techniques has urgency in the development of natural language translation technology and dialogue systems. LLMs often experience obstacles in addressing special requests that require information outside the training data. This study aims to discuss the use of Retrieval-Augmented Generation (RAG) on large-scale language models to improve the performance of Natural Language Processing (NLP) in Indonesian, which has so far been poorly supported by high-quality data and to overcome the limitations of traditional language models in understanding the context of Indonesian better. The method used is a combination of retrieval capabilities (external information search) with generation (text generation), where the model utilizes broader and more structured basic data through the retrieval process to produce more accurate and relevant text. The data used includes the Indonesian corpus of the 30 Juz Quran translation into Indonesian. The results of the trial show that the RAG approach significantly improves the performance of the model in various NLP tasks, including token usage optimization, text classification, and context understanding, by increasing the accuracy and relevance of the result
PERAMALAN PENJUALAN MOBIL DENGAN MENERAPKAN METODE SINGLE MOVING AVERAGE DAN SINGLE EXPONENTIAL SMOOTHING
Get PDFSales forecasting is a critical aspect of the automotive business as it influences marketing strategies, production, and inventory management. This study examines a comparison of Toyota car sales forecasts using the Single Moving Average (SMA) and Single Exponential Smoothing (SES) methods. The primary objective is to evaluate the accuracy of the SES method in predicting Toyota car sales and compare it with actual sales data. Historical sales data serve as the basis for forecasting, with the smoothing parameter α tested to obtain the best results. Forecasting with SMA demonstrates improved accuracy as the number of average periods increases. The 2-period SMA produces forecasts that are more responsive to sales trends, while the 3-period SMA provides more stable predictions without compromising accuracy. Meanwhile, the SES method with α = 0.3 delivers reasonably accurate forecasts. By applying this method, companies can make better decisions regarding Toyota car production and inventory management, ultimately enhancing operational efficiency and customer satisfaction. 
MEASUREMENT OF EMPLOYEE INFORMATION SECURITY AWARENESS: CASE STUDY AT FINANCIAL INSTITUTION
Get PDFThe lack of awareness regarding information security among employees in financial institutions can have detrimental impacts on both customers and the institution itself, both financially and in terms of trust. Therefore, this research aims to assess the information security awareness at PT XYZ, a financial institution, in order to identify the existing level of awareness, which will be used to provide recommendations. The method applied in this research is quantitative, using a questionnaire as a tool for data collection and distribution with a voluntary sampling technique among PT XYZ employees concerning their awareness of information security. The assessment of information security awareness covers 8 specific aspects, with 7 of them drawing sources from HAIS-Q and another 1 referring to the KAMI Index, using the Analytic Hierarchy Process (AHP) for weighting each area. The total number of respondents participating in this research is 52. The research results affirm that PT XYZ employees have a positive awareness of information security, indicating that there are no urgent actions needed at present. However, there are specific areas with potential for improvement, hence recommendations are provided to enhance and sustain information security awareness among employees
DEEP LEARNING FOR POLYCYSTIC OVARIAN SYNDROME CLASSIFICATION USING CONVOLUTIONAL NEURAL NETWORK
Get PDFPolycystic Ovarian Syndrome (PCOS) is the main cause of infertility in women. This condition results in abnormal hormone levels. Women who experience this syndrome will have irregular hormone levels and experience irregular menstrual cycles as well, thereby affecting the reproductive system. Symptoms that arise as a result of the increase in these hormones can be seen from the growth of hair on the legs, weight gain which results in not being ideal, irregular menstruation, unusual acne growth, and oily skin. The problem of Polycystic Ovarian Syndrome can cause disturbances in ovulation and cause infertility in women. Urgency This research requires a classification that has good accuracy in diagnosing early to minimize the rate of pregnancy failure. The aim of the research is to be able to model early detection of Polycystic Ovarian Syndrome with high accuracy so that it can help the health team in detecting Polycystic Ovarian Syndrome or not having Polycystic Ovarian Syndrome. The research stage has 3 stages including the first stage of identifying problems and collecting datasets from Telkom University dataverse in the form of images and literature reviews of various sources. The second stage is Pre Processing of image data, Data Training, modeling design by managing image data and classifying using the Convolutional Neural Network Algorithm deep learning model and testing. The third stage is evaluating the test results and discussing the results of accuracy in determining the status of Normal Polycystic Ovarian Syndrome or PCOS. The results of training and validation on the ovarian xray image dataset using the CNN architecture that has been made, 40 iterations (epochs), and 4 step_per_epochs show an accuracy value of 0.8947 or 89.47% and a loss value of 0.2684