Public Research Journal of Engineering, Data Technology and Computer Science

Application of Recurrent Neural Network Bi-Long Short-Term Memory, Gated Recurrent Unit and Bi-Gated Recurrent Unit for Forecasting Rupiah Against Dollar (USD) Exchange Rate

2024-02-02T02:47:36+00:00

Foreign exchange rates have a crucial role in a country's economic development, influencing long-term investment decisions. This research aims to forecast the exchange rate of Rupiah to the United States Dollar (USD) by using deep learning models of Recurrent Neural Network (RNN) architecture, especially Bi-Long Short-Term Memory (Bi-LSTM), Gated Recurrent Unit (GRU), and Bi-Gated Recurrent Unit (Bi-GRU). Historical daily exchange rate data from January 1, 2013 to November 3, 2023, obtained from Yahoo Finance, was used as the dataset. The model training and evaluation process was performed based on various parameters such as optimizer, batch size, and time step. The best model was identified by minimizing the Mean Square Error (MSE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). Among the models tested, the GRU model with Nadam optimizer, batch size 16, and timestep 30 showed the best performance, with MSE 3741.6999, RMSE 61.1694, MAE 45.6246, and MAPE 0.3054%. The forecast results indicate a strengthening trend of the Rupiah exchange rate against the USD in the next 30 days, which has the potential to be taken into consideration in making investment decisions and shows promising economic growth prospects for Indonesia.

Application of The Fuzzy Mamdani Method in Determining KIP-Kuliah Recipients for New Students

2024-02-02T03:35:12+00:00

Lectures are the last level of education passed. However, the opportunity to obtain further education cannot be owned just like that by everyone because of the economic factors they experience. Therefore, an assessment method is needed to support the decision of KIP-Kuliah recipients at the lecture level for new students within the Faculty of Science and Technology, Sultan Syarif Kasim Riau State Islamic University. This research applies the Fuzzy Mamdani algorithm with Fuzzy Logic and is expected to be able to provide recommendations for worthy scholarship recipients so that the assistance provided is right on target. The results showed that 26,7% of students received the rejected status. Several experiments conducted, illustrate the performance of Fuzzy Logic in this research is very powerful in determining policies and as decision support. The implementation of the research results recommends the best selection from a series of decisions making.

Classifications of Offline Shopping Trends and Patterns with Machine Learning Algorithms

2024-02-02T03:40:57+00:00

Advancements in technology have made online shopping popular among many. However, the use of offline marketing models is still considered a profitable and important way of business development. This can be seen in the 2022 Association of Retail Entrepreneurs of Indonesia (APRINDO), which states that 60% of Indonesians shop offline, and in 2023, more than 75% of continental European consumers will prefer to shop offline. This is because many benefits can be achieved through offline marketing that cannot be obtained from online marketing. Therefore, classification of patterns and trends is performed to compare the results of the algorithms under study. Furthermore, this research was conducted to help offline retailers understand consumption patterns and trends that affect purchases. The algorithms analyzed in this study are K-Nearest Neighbor (K-NN), Naive Bayes, and Artificial Neural Network (ANN). As a result, the ANN algorithm obtained the highest confusion matrix results with an Accuracy value of 96.38%, Precision of 100.00%, and Recall of 100.00%. Meanwhile, when the Naive Bayes algorithm was used, the lowest Accuracy value was 57.39%, the Precision value was 57.86%, and when the K-NN algorithm was used, the Recall value was as low as 92.00%. These results indicate that the ANN algorithm is better at classifying offline shopping image data than the K-NN and Naive Bayes algorithms

Implementation of K-Nearest Neighbors, Naïve Bayes Classifier, Support Vector Machine and Decision Tree Algorithms for Obesity Risk Prediction

2024-01-01T04:13:11+00:00

An abnormal or excessive build-up of fat that can negatively impact one's health as a result of an imbalance in energy between calories consumed and burnt is known as obesity. The majority of ailments, such as diabetes, heart disease, cancer, osteoarthritis, chronic renal disease, stroke, hypertension, and other fatal conditions, are linked to obesity. Information technology has therefore been the subject of several studies aimed at diagnosing and treating obesity. Because there is a wealth of information on obesity, data mining techniques such as the K-Nearest Neighbors (K-NN) algorithm, Naïve Bayes Classifier, Support Vector Machine (SVM), and Decision Tree can be used to classify the data. The 2111 records and 17 characteristics of obesity data that were received from Kaggle will be used in this study. The four algorithms are to be compared in this study. In other words, using the dataset used in this study, the Decision Tree algorithm's accuracy outperforms that of the other three algorithms K-NN, Naïve Bayes, and SVM. Using the Decision Tree algorithm, the accuracy was 84.98%; the K-NN algorithm came in second with an accuracy value of 83.55%; the Naïve Bayes algorithm came in third with an accuracy rate of 77.48%; and the SVM algorithm came in last with the lowest accuracy value in this study, at 77.32%.

Performance Comparison of Random Forest, Support Vector Machine and Neural Network in Health Classification of Stroke Patients

2024-02-02T03:04:22+00:00

Stroke is the second most common cause of death globally, making up about 11% of all deaths from health-related deaths each year, the condition varies from mild to severe, with the potential for permanent or temporary damage, caused by non-traumatic cerebral circulatory disorders. This research began with data understanding through the acquisition of a stroke patient health dataset from Kaggle, consisting of 5110 records. The pre-processing stage involved transforming the data to optimize processing, converting numeric attributes to nominal, and preparing training and test data. The focus then shifted to stroke disease classification using Random Forest, Support Vector Machines, and Neural Networks algorithms. Data processing results from the Kaggle dataset showed high performance, with Random Forest achieving 98.58% accuracy, SVM 94.11%, and Neural Network 95.72%. Although SVM has the highest recall (99.41%), while Random Forest and ANN have high but slightly lower recall rates, 98.58% and 95.72% respectively. Model selection depends on the needs of the application, either focusing on precision, recall, or a balance of both. This research contributes to further understanding of stroke diagnosis and introduces new potential for classifying the disease.

Classification of Diabetes Mellitus Sufferers Eating Patterns Using K-Nearest Neighbors, Naïve Bayes and Decission Tree

2024-01-01T03:18:49+00:00

The study investigates three classification algorithms, namely K-Nearest Neighbor (K-NN), Naïve Bayes, and Decision Tree, for the classification of Diabetes Mellitus using a dataset from Kaggle. K-NN relies on distance calculations between test and training data, using the Euclidean distance formula. The choice of k, representing the nearest neighbor, significantly influences K-NN's effectiveness. Naïve Bayes, a probabilistic method, predicts class probabilities based on past events, and it employs the Gaussian distribution method for continuous data. Decision Trees, form prediction models with easily implementable rules. Data collection involves obtaining a Diabetes Mellitus dataset with eight attributes. Data preprocessing includes cleaning and normalization to minimize inconsistencies and incomplete data. The classification algorithms are applied using the Rapidminer tool, and the results are compared for accuracy. Naïve Bayes yields 77.34% accuracy, K-NN performance depends on the chosen k value, and Decision Trees generate rules for classification. The study provides insights into the strengths and weaknesses of each algorithm for diabetes classification

Evaluation of the Effectiveness of Neural Network Models for Analyzing Customer Review Sentiments on Marketplace

2024-02-02T03:47:45+00:00

According to the 2019 report, Tokopedia is the most visited marketplace with 140,000,000 visitors per month, making it one of the most popular marketplaces in Indonesia. Customers have the opportunity to write reviews about the products they purchase at the end of the transaction process on Tokopedia. The aim of this research is to conduct sentiment analysis on product reviews on Tokopedia. Three neural networks that will be used for text classification are Bi-GRU, GRU, and LSTM. The data processing technique is divided into training and testing samples, split into 80%:20% using the holdout technique. The BI-GRU algorithm has an accuracy of 0.93% and precision of 0.96, better than the other two methods LSTM and GRU, which each have an accuracy of 0.92 and recall of 0.91.