Claim Missing Document
Check
Articles

Found 4 Documents
Search
Journal : Journal of Engineering and Science Application

Harnessing Machine Learning for Stock Price Prediction with Random Forest and Simple Moving Average Techniques Arif Mudi Priyatno; Lidya Ningsih; Muhammad Noor
Journal of Engineering and Science Application Vol. 1 No. 1 (2024): April
Publisher : Institute Of Advanced Knowledge and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.69693/jesa.v1i1.1

Abstract

This paper explores the application of machine learning in predicting stock price trends, specifically for PT Bank Central Asia Tbk (BBCA) shares, using the Random Forest Regression model and Simple Moving Average (SMA) techniques. The SMA parameters ranged from 3 to 200 days, aiding in forecasting the price trends as either rising, sideway, or declining. To achieve accurate and generalizable predictions, the data normalization process was implemented using the MinMax scaler. The methodological framework adopted a time series cross-validation (CV) approach, executed 10 times with a future test window of 40 days, ensuring the robustness and reliability of the predictive model. The model's performance was systematically evaluated based on metrics of accuracy, recall, precision, and F1-score. Results from the cross-validation series indicated varied performance, with the most notable achievements in the 9th and 10th iterations, where both demonstrated an F1-score surpassing 0.745 and 0.808 respectively, and similar levels of accuracy and recall at 0.825. These high F1-scores signify a strong harmonic balance between precision and recall, underscoring the model's capability to effectively predict the stock price movements of BBCA. The findings affirm the potential of utilizing advanced machine learning techniques like Random Forest in conjunction with SMA indicators to enhance the predictability of stock market trends, offering valuable insights for investors and financial analysts.
Comparison of Similarity Methods on New Student Admission Chatbots Using Retrieval-Based Concepts Arif Mudi Priyatno; M. Riko Anshori Prasetya; Putri Cholidhazia; Resy Kumala Sari
Journal of Engineering and Science Application Vol. 1 No. 1 (2024): April
Publisher : Institute Of Advanced Knowledge and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.69693/jesa.v1i1.2

Abstract

A college's students are an essential component. The college always opens registration for new students each year. Every year, more than 1,000 prospective new students register. Because of this, the new student admissions committee is constantly overwhelmed when responding to campus-related questions. As a result, developing a chatbot to assist new students is necessary. The best similarity method is needed for the development of a chatbot using a retrieval-model approach. The New Student Admission Chatbot and the Similarity Method are compared in this study using the Retrieval-Based Concept. The cosine, Jaccard, dice, euclidean, Manhattan, Canberra, and Chebyshev similarity methods are compared. In the context of Universitas Pahlawan Tuanku Tambusai, the data used are information about new students as well as accreditation for study program. There are 41 pieces of information used. Labels and information make up data. According to the test results, the dice and cosine similarity methods are the most effective. On all tested thresholds, dice and cosine similarity achieved an f1-score above 80%. Recall produces extremely optimal results, including 100%.Over 75% of the time, good results are reliably achieved. This demonstrates that the retrieval-model concept can be applied
A Robust Hybrid Approach for Malware Detection: Leveraging CNN and LSTM for Encrypted Traffic Analysis Priyatno, Arif Mudi; Ningsih, Yunia; Vandika, Arnes Yuli; Muhammadong, Muhammadong
Journal of Engineering and Science Application Vol. 1 No. 2 (2024): October
Publisher : Institute Of Advanced Knowledge and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.69693/jesa.v1i2.10

Abstract

The rapid growth in Internet usage and advancements in network technologies have escalated the risk of network attacks. As the adoption of encryption protocols increases, so does the difficulty in identifying malware within encrypted traffic. Malware represents a significant danger in cyberspace, as it compromises personal data and harms computer systems. Network attacks involve unauthorized access to networks, often aiming to disrupt or damage them, with potentially severe consequences. To counter these threats, researchers, developers, and security experts are constantly innovating new malware detection techniques. Recently, deep learning has gained traction in network security and intrusion detection systems (IDSs), with models such as Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) showing promise in detecting malicious traffic. Despite these advancements, extracting relevant features from diverse malware types remains a challenge. Current solutions demand substantial computational resources and are often inefficient for large datasets. Additionally, existing image-based feature extraction methods consume significant resources. This study tackles these issues by employing a 1D CNN alongside LSTM for the detection and classification of encrypted malicious traffic. Using the Malware Analysis benchmark dataset, which consists of 42,797 malware and 1,079 goodware API call sequences, the proposed model achieved an accuracy of 99.2%, surpassing other state-of-the-art models
Evaluating Imputation Approaches and Support Vector Regression Parameters in Weather Forecasting Priyatno, Arif Mudi; Ningsih, Yunia
Journal of Engineering and Science Application Vol. 2 No. 2 (2025): October
Publisher : Institute Of Advanced Knowledge and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.69693/jesa.v2i2.34

Abstract

Rainfall plays a vital role in various sectors such as transportation, agriculture, and industry. Having accurate rainfall information enables stakeholders in these fields to take proper measures and minimize potential losses caused by inaccurate data. This study focuses on identifying an effective method for rainfall forecasting by examining imputation techniques in data preprocessing and parameter settings within Support Vector Regression (SVR). The experimental findings indicate that the most effective imputation method for SVR is determined using the Mean Squared Error (MSE) and Mean Absolute Error (MAE) evaluation metrics. Based on MSE, the k-nearest neighbor method proves to be the most reliable approach for data imputation preprocessing. The preprocessing results were then applied to Polynomial SVR with parameters C = 1000, tolerance = 0.001, epsilon = 0.01, and unlimited iterations. Conversely, MAE results highlight Artificial Neural Network (ANN) as the optimal imputation method. ANN, when combined with a radial basis function kernel, gamma = 0.001, C = 1000, tolerance = 0.001, and unlimited iterations, was further tested using RBF SVR under the same parameter settings.