Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi)
Not a member yet
1071 research outputs found
Sort by
Mapping Residential Land Suitability Using a WEB-GIS-Based Multi-Criteria Spatial Analysis Approach: Integration of AHP and WPM Methods
Get PDFAlong with the increase in population and the acceleration of economic expansion, there has been a concomitant increase in the urgent requirement for additional property that can serve as a venue for a wide variety of community activities. It is not uncommon for large cities, which are the epicenter of urbanization, such as the city of Surabaya, to experience a sharp increase in the demand for land. One of the regions that has excellent accessibility is the Sidoarjo Regency, which is comparable to the City of Surabaya in this regard. The goal of this research is to use Web-GIS to conduct an analysis of spatial data to identify the land functions that are most suitable for use in residential areas. The Analytic Hierarchy Process (AHP) and the Weighted Product Model (WPM) are two of the methodologies that are included in the spatial data modeling method that uses multi-criteria decision making (MCDM). The parameters of the characteristics that are used are derived from data such as the distance to the city center, the distance to the market, the distance to the hospital, the distance to public transportation, the slope, the type of soil, and the amount of rainfall. The results of the spatial data modeling categorize the suitability of new residential land into categories of land that is not suitable for residential use and land that is acceptable for residential use. A K value of 0.27 is the result of the comparison test that was run between the two MCDM approaches using Cohen's Kappa coefficients.Seiring dengan peningkatan jumlah penduduk dan percepatan pertumbuhan ekonomi, kebutuhan mendesak akan tambahan properti yang dapat berfungsi sebagai tempat berbagai kegiatan masyarakat juga meningkat. Tidak jarang kota-kota besar yang menjadi episentrum urbanisasi, seperti kota Surabaya, mengalami lonjakan permintaan lahan yang cukup tajam. Salah satu daerah yang memiliki aksesibilitas yang sangat baik adalah Kabupaten Sidarjo yang sebanding dengan Kota Surabaya dalam hal ini. Tujuan dari penelitian ini adalah menggunakan Web-GIS untuk melakukan analisis data spasial guna mengidentifikasi fungsi lahan yang paling sesuai untuk digunakan di kawasan pemukiman. Analytic Hierarchy Process (AHP) dan Weighted Product Model (WPM) merupakan dua metodologi yang termasuk dalam metode pemodelan data spasial yang menggunakan multi-criteria decision making (MCDM). Karakteristik parameter yang digunakan diperoleh dari data seperti jarak ke pusat kota, jarak ke pasar, jarak ke rumah sakit, jarak ke angkutan umum, kemiringan lereng, jenis tanah, dan jumlah curah hujan. Hasil pemodelan data spasial mengkategorikan kesesuaian lahan pemukiman baru ke dalam kategori lahan tidak layak huni dan lahan layak huni. Nilai K sebesar 0.27 merupakan hasil uji perbandingan yang dijalankan antara kedua pendekatan MCDM menggunakan koefisien Cohen's Kappa
Anatomy Identification of Bamboo Stems with The Convolutional Neural Networks (CNN) Method
Get PDFIt is important to note that some species of bamboo are protected and considered endangered. However, distinguishing between traded and protected bamboo species or differentiating between bamboo species for various purposes remains a challenge. This requires specialized skills to identify the type of bamboo, and currently, the process can only be carried out in the forest for bamboo that is still in clump form by experienced researchers or officers. However, a study has been conducted to develop an easier and faster method of identifying bamboo species. The study aims to create an automatic identification system for bamboo stems based on their anatomical structure (ASINABU). The bamboo identification algorithm was developed using macroscopic images of cross-sectioned bamboo stems and the research method used was the convolutional neural network (CNN). CNN was designed to identify bamboo species with images taken using a cellphone camera equipped with a lens. The final product is an Android automatic identification application that can detect bamboo species with an accuracy of 99.9%.Beberapa spesies bambu dilindungi dan dikategorikan terancam punah. Bagaimana membedakan jenis bambu yang diperdagangkan dan yang dilindungi, atau bagaimana membedakan jenis bambu untuk berbagai kegunaan, masih belum dikuasai. Hal ini membutuhkan keterampilan untuk mengidentifikasi jenis bambu. Saat ini proses identifikasi bambu hanya dapat dilakukan di hutan untuk bambu yang masih berupa rumpun oleh peneliti atau petugas yang terampil, terlatih dan berpengalaman, dimana kondisi individu menentukan hasil dan lamanya waktu identifikasi. Penelitian ini bertujuan untuk mengembangkan sistem identifikasi otomatis batang bambu berdasarkan struktur anatominya (ASINABU), sehingga lebih mudah dalam mengidentifikasi spesies bambu dengan cepat. Algoritma identifikasi bambu dikembangkan berdasarkan karakteristik khusus dari pola anatomi potongan melintang batang bambu menggunakan struktur citra makro. Metode penelitian yang digunakan adalah Convolutional Neural Network (CNN) yang dirancang untuk dapat mengidentifikasi spesies bambu dengan citra makroskopik yang diambil menggunakan kamera pada handphone yang dilengkapi dengan lensa. Proses terakhir berupa aplikasi pengenal otomatis di Android. Hasil penelitian menunjukkan bahwa aplikasi ASINABU dapat mendeteksi spesies bambu dengan akurasi hingga 99%
Perbandingan Algoritma Genetika dan Recursive Feature Elimination pada High Dimensional Data
No full textThe use of big data in companies is currently used in file processing. With large capacity files, it can affect the performance in terms of time in the company, so to overcome the problem of high-dimensional data, feature selection is used in selecting the number of features. On the WDC dataset with 30 features and 569 data points, feature selection is performed using the Recusive Feature Elimination (RFE) and Genetic Algorithm (GA) models. Then, a comparison of evaluation values is made to determine which feature selection is best for solving the problem. From the 14 tables of evaluation results and discussion in tables 1 to 14, it is found that in the evaluation of accuracy and the use of weighted macros on precision, recall, and f1 score, using GA selection features has slightly higher results than RFE, so it is concluded that GA selection features are better at solving problems in high-dimensional data.Penggunaan big data di perusahaan saat ini digunakan dalam pengolahan file. Dengan file yang berkapasitas besar dapat mempengaruhi kinerja dari segi waktu pada perusahaan sehingga untuk mengatasi masalah data yang berdimensi tinggi maka digunakan feature selection dalam pemilihan jumlah fitur. Pada dataset wdbc dengan total 30 fitur dengan total 569 data, seleksi fitur dilakukan dengan menggunakan model Recusive Feature Elimination (RFE) dan Genetic Algorithm (GA). Kemudian dilakukan perbandingan nilai evaluasi untuk menentukan pemilihan fitur mana yang terbaik untuk menyelesaikan masalah. Dari 16 tabel hasil evaluasi dan pembahasan pada tabel 1 sampai dengan 16, diketahui bahwa fitur pemilihan GA sangat mendominasi untuk digunakan pada permasalahan data dimensi tinggi, hal ini didasarkan pada akurasi, presisi, recall dan f-score. menggunakan model weighted dan unweighted (macro), hampir semua nilai evaluasi dengan fitur seleksi GA lebih tinggi dari fitur seleksi RFE
MRI Image Based Alzheimer’s Disease Classification Using Convolutional Neural Network: EfficientNet Architecture
Get PDFAlzheimer's disease is a neurodegenerative disorder or a condition characterized by degeneration and damage to the nervous system. This leads to a decline in cognitive abilities such as memory, thinking, and focus, which can impact daily activities. In the medical field, a technology called Magnetic Resonance Imaging (MRI) can be used for the initial diagnosis of Alzheimer's disease through image procedures-based recognition methods. The development of this detection system aims to assist medical professionals, including doctors and radiologists, in diagnosing, treating, and monitoring patients with Alzheimer's disease. This study also aims to classify different types of Alzheimer's disease into four distinct classes using the convolutional neural network method with the EfficientNet-B0 and EfficientNet-B3 architectures. This study used 6400 images that encompass four classes, namely mild demented, moderate demented, non-demented, and very mild demented. After conducting testing for both scenarios, the exactness outcomes for scenario 1 utilizing EfficientNet-B0 reveryed 96.00%, and for scenario 2 utilizing EfficientNet-B3, the exactness was 97.00%
Increasing the Accuracy of Brain Stroke Classification using Random Forest Algorithm with Mutual Information Feature Selection
No full textBrain stroke stands out as a leading cause of death, distinguishing it from common illnesses and highlighting the critical need to utilize machine learning techniques to identify symptoms. Among these techniques, the Random Forest (RF) algorithm emerged as the main candidate because of its optimal accuracy values. RF was chosen for its ensemble learning properties that optimize accuracy while simultaneously, bagging all outputs (DT), thus increasing its efficacy. Feature Selection, an important data analysis step, which is mainly achieved through pre-processing, aims to identify influential features and ignore less impactful features. Mutual Information serves as an important feature selection method. Specifically, the highest level of accuracy was achieved by cross-validating the test data - 10, resulting in 0.7760 without feature selection and 0.7790 with mutual information. Most of the attributes in the brain stroke dataset show relevance to the stroke disease class, but the resulting decision tree shows age as a particularly important node. So, the research results show that the selection feature (Mutual Information) can increase the accuracy of brain stroke classification, although it is not significant, namely an increase of 0.0030%. With an increase, where there is no significant difference, it can be said that almost all the attributes contained in the brain stroke dataset used have an influence on their relevance to the stroke disease class
Improving Diabetes Prediction Accuracy in Indonesia: A Comparative Analysis of SVM, Logistic Regression, and Naive Bayes with SMOTE and ADASYN
Get PDFThis study aims to enhance the accuracy of diabetes prediction models in Indonesia by comparing the performance of Support Vector Machines (SVM), Logistic Regression, and Naïve Bayes algorithms, both with and without synthetic oversampling techniques such as SMOTE and ADASYN. The research addresses the issue of imbalanced datasets in medical diagnostics, specifically in predicting diabetes among Indonesian patients, where such imbalance often leads to biased predictions. A comprehensive dataset comprising 657 patient records from a Regional General Hospital in Indonesia was used, with 70% of the data allocated for training and 30% for testing. The results indicate that the SVM model combined with SMOTE achieved the highest accuracy of 95.8% and an AUC of 99.1, underscoring the effectiveness of these techniques in improving prediction performance. The findings of this study highlight the importance of selecting appropriate oversampling methods and algorithms to optimize diabetes prediction accuracy in the Indonesian context, providing valuable insights for future healthcare strategies
Cross-Spectral Cross-Distance Face Recognition via CNN with Image Augmentation Techniques
Get PDFFacial recognition is a critical biometric identification method in modern security systems, yet it faces significant challenges under varying lighting conditions, particularly when dealing with near-infrared (NIR) images, which exhibit reduced illumination compared to visible light (VIS) images. This study aims to evaluate the performance of Convolutional Neural Networks (CNNs) in addressing the Cross-Spectral Cross-Distance (CSCD) challenge, which involves face identification across different spectra (NIR and VIS) and varying distances. Three CNN models—VGG16, ResNet50, and EfficientNetB0—were assessed using a dataset comprising 800 facial images from 100 individuals, captured at four different distances (1m, 60m, 100m, and 150m) and across two wavelengths (NIR and VIS). The Multi-task Cascaded Convolutional Networks (MTCNN) algorithm was employed for face detection, followed by image preprocessing steps including resizing to 224x224 pixels, normalization, and homomorphic filtering. Two distinct data augmentation strategies were applied: one utilizing 10 different augmentation techniques and the other with 4 techniques, trained with a batch size of 32 over 100 epochs. Among the tested models, VGG16 demonstrated superior performance, achieving 100% accuracy in both training and validation phases, with a training loss of 0.55 and a validation loss of 0.612. These findings underscore the robustness of VGG16 in effectively adapting to the CSCD setting and managing variations in both lighting and distance
Machine Learning Methods for Forecasting Intermittent Tin Ore Production
Get PDFEffective production forecasting is important for resource planning and management in the mining industry. Tin ore production from Cutter Section Dredges (CSD) may fluctuate due to a variety of factors, in which there are periods when the production is zero. This study compares various combinations of machine learning-based classification and forecasting to predict future tin ore production values, which have not been found in previous studies. The presence of zero values in the forecast in the next day's tin ore production forecast is addressed by combining classification and forecasting techniques. Random Forest and CatBoost classification techniques are used to determine the next day's CSD production operating status. Then, for each time point when the CSD is operational, a forecasting model is created using CatBoost and Bi-LSTM. This study's findings show that a serial combination of the Random Forest classification method and CatBoost forecasting can produce accurate tin ore production forecasts for the selected CSD (RMSE = 0.271, MAE = 0.179, MAE = 0.730, F1-score = 0,80). This study demonstrates how a serial combination of classification and forecasting models can improve the accuracy and efficiency of production forecasting for intermittent time series data
Efficient Pattern Recognition of Sundanese Script Variants Using CNN
No full textThis research aims to apply pattern recognition technology, specifically through the Convolutional Neural Network (CNN) approach, in identifying and translating Sundanese script accurately. This research is focused on recognizing rarangken script patterns based on ngalagena script in Indonesian cultural heritage. This study uses the MobileNetV2 based CNN model, utilizing transfer learning and trained for 50 epochs using the Adam optimizer with a learning rate of 0.0001, to achieve a training accuracy of 98.75% and test accuracy of 96.95% in 1 hour and 23 minutes, respectively. The results of the study show that the simpler CNN architecture without augmentation achieved the highest accuracy of 99.26%, and the augmented CNN model achieved 94.42% accuracy in 2 hours and 22 minutes. These results enable practical applications in both education and cultural preservation, demonstrating how modern technology can effectively contribute to maintaining traditional cultural elements in the digital era
Cattle Weight Estimation Using Linear Regression and Random Forest Regressor
Get PDFThe global cattle farming industry has benefits as a food source, livelihood, economic contribution, land environmental restoration, and energy source. The importance of predicting cow weight for farmers is to monitor animal development. Meanwhile, for traders, knowing the animal's weight makes it easier to calculate the price of the animal meat they buy. The authors propose estimating cattle weighting linear regression and random forest regression. Linear regression can interpret the linear relationship between dependent and independent variables, and random forest regression can generalize the data well. The data set used in this study consisted of ten variables: live body weight, withers height, sacrum height, chest depth, chest width, maclocks width, hip joint width, oblique body length, oblique back length and chest circumference. Find the model that produces the smallest MAE value. The results show that the linear regression algorithm can produce estimated weight values for cattle with the best performance. This model produces a mean absolute error (MAE) of 0.35 kg, a mean absolute percentage error (MAPE) of 0.07%, a root mean square error (RMSE) of 0.5 kg, and an R² of 0.99. Each variable has excellent correlation performance results and contributes to computer vision and machine learning