ejournal.nusamandiri.ac.id (STMIK Nusa Mandiri)
Not a member yet
1504 research outputs found
Sort by
IMPROVING AGRICULTURAL YIELDS IN THE DEMOCRATIC REPUBLIC OF CONGO USING MACHINE LEARNING ALGORITHMS
Get PDFThis article presents an analysis of agricultural yields in the Democratic Republic of Congo (DRC) using machine learning algorithms. The study is based on around 30,000 records covering several years of agricultural production. Each record includes variables such as seed type, climatic conditions (temperature, rainfall and humidity), soil characteristics (pH, nutrients), farming practices (fertilizer use, irrigation) and yields obtained. The data comes from a variety of sources, including METTELSAT, the World Meteorological Organization (WMO) and WorldClim for climate data, and the DRC Ministry of Agriculture and the FAO for soil and agricultural data. The algorithms evaluated include linear regression, random forest regression, Gradient Boosting Machines (GBM), Support Vector Machines (SVM), and Artificial Neural Networks (ANN). The performance of the algorithms is measured using metrics such as MSE, MAE, RMSE, R² Score and MAPE on three separate case studies (Farm A, Farm B and Farm C). The results show that artificial neural networks (ANNs) perform best, with MSE ranging from 600 to 850, MAE from 12 to 17, RMSE from 24.49 to 29.15, R² Score from 0.92 to 0.95, and MAPE from 8.5% to 10.7%. Next came GBM, random forest regression, SVM and finally linear regression. These results highlight the potential of machine learning algorithms to improve agricultural yield forecasts in the DRC
COMPARATIVE ANALYSIS OF RANDOM FOREST AND SUPPORT VECTOR CLASSIFIER FOR PREDICTING STUDENTS’ ON-TIME GRADUATION
Get PDFOn-time graduation is one of the key indicators of educational quality in higher education. The influencing factors range from students’ internal issues and academic abilities to institutional policies. However, academic management has not yet been able to classify the data and analyze the underlying factors contributing to delayed graduation. By identifying these factors, management can formulate appropriate academic solutions or policies. The purpose of this study is to build a prediction model for on-time graduation using machine learning algorithms. This study compares the classification performance of the Random Forest algorithm and the Support Vector Classifier (SVC). The dataset, consisting of 1,298 student records, includes academic data such as study program, GPA, TOEFL score, cohort year, and study duration. Model performance was evaluated using accuracy, F1 score, and ROC-AUC metrics, followed by a confusion matrix analysis. The final evaluation revealed that the Random Forest algorithm achieved the best performance, with an accuracy of 91.86%, an F1 score of 91.86%, and a ROC-AUC of 97.39%. Meanwhile, the SVC model obtained an accuracy of 81.12% and an F1 score of 81.09%. Based on these results, it can be concluded that the Random Forest algorithm is more reliable as a prediction model in the academic domain. The main contribution of this study is the development of an early detection system for students at risk of delayed graduation. Furthermore, the findings can serve as a basis for designing more solution-oriented academic policies in accordance with the conditions at STIMIK Tunas Bangsa Banjarnegara
PARKINSONS DISEASE DETECTION USING INCEPTION AND X-CEPTION WITH ATTENTION MECHANISM
Get PDFParkinson's disease is one of the global health challenges that requires early detection to slow the progression of symptoms. This study proposes an automatic detection system based on deep learning using the InceptionV3 and Xception architectures combined with a multi-head awareness mechanism. The dataset used consists of 72 handwritten spiral images, comprehensively distributed between the Healthy and Parkinson's categories. The process includes preprocessing in the form of normalization and image resizing, as well as model training using the Adam algorithm and the binary cross-entropy loss function. The results show that the model is able to classify both categories with high accuracy. The use of the attention mechanism provides a performance increase of 4.2% on InceptionV3 and 3.1% on Xception compared to the version without attention. In data testing, the InceptionV3 model with attention achieved 100% accuracy, 100% precision, 100% recall, and 100% F1-score. Meanwhile, the Xception model with attention achieved 88% accuracy, 90% precision, 88% recall, and 87% F1-score. The attention mechanism also helps the model in capturing important features such as vibration and irregularity of the spiral pattern. This research makes an important contribution to the development of an artificial intelligence-based automated early diagnosis system to detect Parkinson's disease more accurately and responsively
GFPGAN UPSCALING FOR HUMAN FACIAL EXPRESSION CLASSIFICATION USING VGG19 ARCHITECTURE
Get PDFHuman facial expression recognition is a rapidly evolving field in artificial intelligence and digital image processing. This study aims to develop a model capable of recognizing and classifying human emotions through facial feature analysis. However, a major challenge in facial expression classification is low image quality, which can reduce model accuracy. Factors such as poor lighting, low resolution, variations in viewing angles, and occlusion (obstructions) on the face pose significant obstacles to accurate detection.This research proposes the application of an upscaling method using the Generative Facial Prior Generative Adversarial Network (GFPGAN) to enhance facial image quality by restoring details in expressions that may be unclear due to low resolution. After the upscaling process, facial expression classification is conducted using a CNN architecture based on VGG19, and the model is evaluated using accuracy, precision, recall, and F1-score metrics to assess its performance in emotion detection. Experiments are conducted in two scenarios: classification without upscaling and classification with GFPGAN upscaling. The results indicate that integrating GFPGAN with the VGG19-based CNN proposed in this study significantly improves emotion detection accuracy, achieving 86%, compared to 76% for the model without image quality enhancemen
OPTIMIZING MULTI-CHANNEL RESNET50 FOR CITRUS LEAF CLASSIFICATION USING COLOR ENHANCEMENT AND EDGE DETECTION METHOD
Get PDFConventional methods face limitations due to the high similarity in color and morphology among citrus leaves classification. To address this challenge, deep learning approaches combined with advanced image preprocessing techniques offer a promising solution. This study employed transfer learning using the ResNet50 architecture integrated with image preprocessing methods including contrast enhancement and edge detection. The experiment was implemented in Python 3.13.2 with TensorFlow on an HP OMEN laptop equipped with Intel® Core™ i7-12700F and NVIDIA® GeForce RTX™ 3060 Ti GPU. A dataset of 250 images across five citrus species was captured using a Samsung M54 camera. To enhance dataset diversity, augmentation techniques such as zoom scaling (80–120%), random rotation (±15° to +30°), and horizontal/vertical translation (10–20%) were applied, expanding the dataset to 2,500 images. Data were divided into training (70%), validation (15%), and testing (15%). Four model scenarios were evaluated: MSR-ResNet50 (RGB), GC-ResNet50 (RGB), LF-ResNet50 (GS), and GC-MSR-LF MC-ResNet50 (RGB+GS). Among the evaluated models, GC-MSR-LF MC-ResNet50 achieved the best performance, recording accuracies of 93.7% for training, 91.0% for validation, and 90.2% for the test set. These results indicate a significant improvement in distinguishing citrus leaves with high morphological similarity. The findings confirm that integrating image preprocessing methods with transfer learning enhances the accuracy of citrus leaf classification. The proposed GC-MSR-LF MC-ResNet50 model demonstrates robust generalization across datasets, highlighting its potential application in precision agriculture for automated species identification and crop monitoring
ANALISIS SENTIMEN APLIKASI TIKTOK SHOP SELLER CENTER MENGGUNAKAN NAIVE BAYES, SVM DAN LOGISTIC REGRESSION
Get PDFThe rapid growth of e-commerce has driven the emergence of new platforms such as TikTok Shop Seller Center, which is now integrated with Tokopedia. Increasing competition among digital platforms has made service quality and user experience key success factors. In this context, user reviews and feedback serve as crucial data sources that reflect satisfaction, complaints, and expectations toward the application. However, the large and diverse volume of reviews renders manual analysis inefficient. Therefore, an automated approach such as sentiment analysis is required to classify user opinions quickly and accurately. This study aims to perform sentiment analysis on TikTok Shop Seller Center user reviews using Naïve Bayes, Support Vector Machine (SVM), and Logistic Regression algorithms to determine the best-performing model. The dataset was obtained from the Kaggle platform and underwent preprocessing, including case folding, tokenization, stemming, and TF-IDF weighting. Model evaluation was conducted using confusion matrix and ROC curve, along with performance metrics such as accuracy, precision, recall, and F1-score. The results show that the SVM algorithm outperformed Naïve Bayes and Logistic Regression, achieving 93.75% accuracy, 93.78% precision, 95.65% recall, 94.70% F1-score, and an AUC of 0.98, categorized as Excellent Classification. Thus, SVM proved to be the most effective algorithm for classifying user review sentiments on TikTok Shop Seller Center
COMPARATIVE PERFORMANCE OF TRANSFORMER AND LSTM MODELS FOR INDONESIAN INFORMATION RETRIEVAL WITH INDOBERT
Get PDFNeural network-based Information Retrieval (IR), particularly with Transformer models, has gained prominence in information search technology. However, the application of this technology in Indonesian, a low-resource language, remains limited. This study aims to compare the performance of the LSTM model and IndoBERT for IR tasks in Indonesian. The dataset consists of 5,000 query–document pairs collected via scraping from three Indonesian news portals: CNN Indonesia, Kompas, and Detik. Evaluation was performed using MAP, MRR, Precision@5, and Recall@5 metrics. The results show that IndoBERT outperforms LSTM in all metrics with a MAP of 0.82 and MRR of 0.84, while LSTM only reached a MAP of 0.63 and MRR of 0.65. These findings confirm that Transformer models like IndoBERT are more effective at capturing semantic relevance between queries and documents, even with limited datasets
CLASSIFICATION OF PAPAYA NUTRITION BASED ON MATURITY WITH DIGITAL IMAGE AND ARTIFICIAL NEURAL NETWORK
Get PDFPapaya is a tropical fruit with high nutritional content and significant health benefits. Nutritional components such as sugars, vitamin C, and fibre are strongly influenced by ripeness level. Identifying these nutrients usually requires laboratory tests that are time-consuming and rely on sophisticated equipment. Previous studies have focused on classifying ripeness levels, yet none have specifically addressed the classification of nutritional content. This study proposes a classification system for papaya nutrition based on ripeness using digital image processing and artificial neural networks (ANN). The method consists of six stages: image acquisition, preprocessing, segmentation, morphology, feature extraction, and classification with a trained ANN model. Experiments were conducted to evaluate feature combinations, including colour and texture features. The combination of LAB colour features and texture features-contrast, correlation, energy, and homogeneity-produced the best results. Testing on 75 images achieved an average precision of 97.22%, recall of 96.67%, F1-Score of 96.80%, and accuracy of 97.33%, with an average computation time of 0.02 seconds per image. These findings indicate that the proposed method provides fast and highly accurate classification of papaya’s nutritional content, offering a practical alternative to laboratory testing. Nevertheless, the study is limited by the relatively small dataset and controlled acquisition environment. Future research should extend the dataset, incorporate deep learning approaches, and validate performance under real-world conditions to enhance robustness and generalizatio
PREDICTIVE MODEL FOR COOPERATIVE LOAN RECIPIENT ELIGIBILITY USING SUPERVISED MACHINE LEARNING
Get PDFNon-performing loans remain a critical challenge for cooperatives as they can undermine financial stability, erode member trust, and impede institutional growth. This study develops a predictive model for cooperative loan eligibility using supervised machine learning techniques and a novel three-class classification framework, Approved, Consideration, and Rejected, to support more objective and transparent decision-making. A dataset of 1,000 borrower records containing demographic and financial attributes was analyzed using Naive Bayes, Decision Tree, and Random Forest algorithms implemented in RapidMiner. The Random Forest algorithm achieved the best predictive performance with an accuracy of 96.02%, demonstrating its robustness and reliability compared to the other models. The proposed three-class system differentiates this study from conventional binary classification approaches, enabling finer distinctions among borrower categories and promoting fairness in cooperative credit evaluations. The findings provide practical guidance for cooperatives to adopt data-driven, transparent, and accountable decision-making systems that reduce manual bias and strengthen financial inclusion. Overall, the proposed three-class model built through a supervised learning framework offers a reliable, fair, and scalable solution to support sustainable lending practices and enhance risk management in cooperative institutions
PENGARUH CITRA WISATA DAN DAYA TARIK TERHADAP KEPUTUSAN BERKUNJUNG PADA OBJEK WISATA VELANGKANNI MEDAN
Get PDFGraha Maria Annai Velangkanni plays an important role as an icon of Medan City as well as a tourist destination that attracts visitors due to its historical value and beauty, contributing to the development of the local tourism industry. From January to April 2025, the number of tourist visits to this site fluctuated, which became the main focus of this study. Factors influencing tourists’ decisions to visit include tourism image and tourist attractions, considered key aspects in visitor preferences. This study aims to analyze the effect of tourism image and tourist attractions on the decision to visit Graha Maria Annai Velangkanni in Medan. The research employed a quantitative method, collecting primary data through survey questionnaires and literature review. Data analysis included validity and reliability tests, classical assumption tests, hypothesis testing using multiple linear regression, t-tests, F-tests, and calculation of the coefficient of determination (Adjusted R²) with a sample of 100 respondents. The results of the multiple linear regression showed a positive effect, with regression coefficients of 0.315 for tourism image and 0.845 for tourist attractions. The t-test results indicated t-values of 2.884 (p=0.005) for tourism image and 12.932 (p=0.001) for tourist attractions. The F-test produced an F-value of 100.429 with a significance of 0.001. These findings confirm that tourism image and tourist attractions have a positive and significant influence, both partially and simultaneously, on tourists’ decisions to visit Graha Maria Annai Velangkanni in Medan.Faktor citra wisata dan daya tarik wisata merupakan 2 dari banyak faktor yang mempengaruhi keputusan berkunjung pada objek wisata Graha Maria Annai Velangkanni. Penelitian ini bertujuan untuk mengukur pengaruh citra wisata dan daya tarik wisata terhadap keputusan berkunjung pada objek wisata Graha Maria Annai Velangkanni Medan. Metode penelitian ini menggunakan pendekatan kuantitatif dengan pengumpulan data primer melalui kuesioner dan studi pustaka. Berdasarkan hasil penelitian pada taraf signifikansi 5% menunjukkan bahwa citra wisata dan daya tarik wisata berpengaruh positif dan signifikan secara parsial maupun simultan terhadap keputusan berkunjung pada objek wisata Graha Maria Annai Velangkanni. Kemudian variabel citra wisata dan daya tarik wisata, secara bersama-sama atau simultan berpengaruh signifikan terhadap keputusan berkunjung sebesar 100,429 lebih besar dari ftabel 3,09