Journal of Applied Computer Science and Technology

Peramalan Harga Cabai Rawit Merah Menggunakan Attention Mechanism Berbasis Long Short-Term Memory

2025-01-01T01:32:52+00:00

Red cayenne pepper is a commodity that has important economic value in Indonesia, especially in West Java Province. Cayenne pepper often experiences significant price fluctuations which can cause inflation. In 2022 there will be the highest inflation in West Java in the last eight years due to increased commodity prices, including cayenne pepper. There needs to be an effort to maintain the stability of the price of red cayenne pepper in West Java. This research aims to create a price forecasting system for red cayenne peppers in West Java by comparing two deep learning approaches, namely Long-Short Term Memory (LSTM) and Long-Short Term Memory with Attention Mechanism (LSTM-Attention-LSTM) to obtain high accuracy in predicting the price of red cayenne pepper. The results of this research show that the LSTM model using 3 hidden layers, 100 neurons, 128 dense, 1 dense, and 32 batch sizes, produces Mean Absolute Error (MAE) values of 0.023, Root Mean Square Error (RMSE) of 0.152, and Mean Absolute Percentage Error (MAPE) is 3.68%. Meanwhile, the LSTM-Attention-LSTM model with the same configuration produces an MAE value of 0.017, RMSE of 0.130, and MAPE of 2.74%. The results of this research can be a reference for the community and government in maintaining price stability for cayenne pepper in West Java.

Identifikasi Kelayakan Air Minum Dengan Metode Analisis Komponen Utama Berbasis Entropi

2025-01-01T01:32:39+00:00

The need for clean water is a fundamental requirement that must be met by humans, as water constitutes 60 to 70% of the total human body weight. Therefore, it is important to be able to determine the quality of the water entering the body, as consuming unsafe water will bring various diseases, such as diarrhea, and in severe cases might lead to death. This study aimed to investigate the factors which determine the potability of drinking water. Specifically, this research aims to produce a fault detection algorithm that can detect the potability of water samples based on Principal Component Analysis (PCA) and entropy-based subset selection methods. This paper addresses the linearity problem that commonly occurred in PCA by finding a subset of data that has a good entropy relation among the parameters contained in the subset, thus maintaining linearity in the data. There were 8 parameters considered in this reseach: pH, hardness, total dissolved solids, chloramines, sulfate, conductivity, organics carbon, trihalomethanes and turbidity. The experiment was conducted with 811 water samples, where 645 samples were used to train the model and the rest for validating the model predictive accuracy. Based on experiments conducted, it is confirmed that the proposed algorithm can determine the potability of drinking water samples from synthetic data sourced from India with an accuracy of over 98% for potable water data and 100% for non-potable water data.

Klasifikasi Multilabel Pada Gaya Belajar Siswa Sekolah Dasar Menggunakan Algoritma Machine Learning

2025-01-01T01:32:24+00:00

Learning style plays a very important role in determining the success of a person's learning process. An individual generally has a combination of all three existing learning styles including Visual, Auditorial, and Kinesthetic. However, what distinguishes the abilities of individuals from each other is how the dominant combination of each learning style is or not, so it is important to identify. This study aims to build a multi-label classification model to classify the learning styles of elementary school students. The machine learning algorithms used to build the model are Decision Tree, K-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Multi-Layer Perceptron (MLP). The performance of these four models is compared using the Hamming Loss, Accuracy, Precision, Recall, and F1-Score performance metrics. The Classifier Chains method is implemented to provide capabilities to KNNs and SVMs that cannot directly handle multi-label classification problems. The dataset used in this study is the Data Set of Learning Style Preference. The separation of the dataset was made into three different forms of data sizes, including: Data I: 90% training, 10% testing; Data II: 80% training 20% testing; and Data III: 70% training 30% testing. Cross-validation using K-Fold Cross Validation with a k-value of 10-fold was also applied to the training data. Based on testing, the best performance was obtained on the Decision Tree model with a hamming loss of 0.014, which indicates a very low prediction error rate per individual label. A recall value of 99% indicates that the model is able to detect almost all positive labels correctly, and an F1-score of 98% indicates that the model has excellent and balanced performance, without bias against both positive and negative label predictions. The performance of the Decision Tree model was followed by MLP, SVM, and KNN which showed lower results.

Klasifikasi Kualitas Biji Kedelai Menggunakan Transfer Learning Convolutional Neural Network Dan SMOTE

2025-01-01T01:32:11+00:00

Soybeans are an important commodity in the food and feed industry, where they can be used to produce soy milk or other processed products. However, low-quality seeds can reduce the quality of the processed products and increase production costs. To address this issue, a soybean seed quality classification system was developed using the method of Transfer Learning with CNN and SMOTE. This method leverages the ability of neural networks for extracting of the visual features and handle for data imbalance between classes. The study shows that the CNN model achieved value accuracy is 91.09%, while the combination of CNN and SMOTE get the accuracyf 89.92%. Additionally, the MobileNetV2 model reached an accuracy of 91.11%, which further improved to 92.42% after applying SMOTE. These results demonstrate that the use of Transfer Learning and SMOTE significantly enhances accuracy in soybean seed quality classification, resulting in a more effective.

Evaluasi Kerentanan Insecure Direct Object Reference pada Aplikasi Pendaftaran Sidang Universitas XYZ

2025-01-01T01:33:06+00:00

This study aims to analyze and evaluate the vulnerability of Insecure Direct Object Reference (IDOR) in the thesis registration web application at XYZ University, as well as to provide improvement recommendations to enhance the security of students' personal data. The IDOR vulnerability allows unauthorized access to students' personal documents, which can jeopardize privacy and information security. Utilizing an action research methodology consisting of four stages: diagnosis, action taking, evaluation, and learning, this research identifies the URL patterns generated when students upload documents such as ID cards, family cards, birth certificates, diplomas, and photos. During the action-taking phase, the researcher conducts attacks using Burp Suite to test the vulnerability by modifying URL parameters based on the identified patterns. The testing results indicate that all documents can be accessed without proper authorization, with a status code of 200 indicating successful access. These findings underscore the necessity for stricter security improvement measures in the thesis registration application to protect students' personal data. The implications of this research highlight the importance of implementing tighter access controls and better input validation in higher education applications to prevent potential data leaks in the future. This study makes a significant contribution to enhancing information security within educational environments.

Implementasi Keamanan Ruangan Berbasis IoT dengan Sensor PIR, Telegram, dan Peringatan Suara

2025-01-01T01:31:56+00:00

Home security is crucial for protecting property and occupant safety. Conventional technology often falls short in providing optimal protection. With advancements in the Internet of Things (IoT), smarter and more affordable security solutions are now accessible. This research develops an IoT-based room security system using a PIR sensor to detect motion, Telegram notifications for real-time communication, and sound alerts as preventive measures. The NodeMCU ESP8266 module processes data from the PIR sensor and transmits signals via WiFi. Test results show that the PIR sensor successfully detects motion up to 5 meters with a 93.33% success rate. The system also effectively sends real-time notifications to the Telegram app and provides efficient sound alerts. Thus, this system significantly enhances home security with minimal cost and complexity, offering convenience and quick response to users.

Implementasi Algoritma Support Vector Machine (SVM) Untuk Klasifikasi Penyakit Stroke

2025-01-01T01:31:40+00:00

Stroke or known as Cerebrovascular Accident (CVA) is a functional disorder caused by impaired blood flow function from within the human brain. Stroke carries a high risk of brain damage, paralysis, speech disorders, visual impairment, even death. Stroke is one of 10 life-threatening diseases in Indonesia. Classification is one of a few methods in predicting stroke symptoms with the aim of obtaining accurate prediction of disease. The researchers implemented a method to classify stroke with the Support Vector Machine (SVM) algorithm. The SVM is a learning method used in medical diagnosis for classification, the researchers processed data sets using the Orange tool. The research applied cross validation techniques to improve the accuracy of the model. The study used data sets from the data.world.com site with a total of 40,910 data consisting of 11 attributes. In this process, 80% of the training data and 20% of the test data are used. Using the Orange tool, the study managed to classify stroke disease well using the RBF kernel with cross validation techniques resulting in an accuracy of 94.8%. The results of this study can be concluded that the stroke classification model developed has excellent performance. Overall, these results indicate that the Stroke classification model developed is highly reliable and effective, with excellent ability to detect stroke cases and provide accurate predictions. Making better and quicker medical judgments can be aided by using this approach to diagnose strokes.