cover
Article Per Year (5 Year)
All
Home Page
OAI Link
Editorial Team
Contact
Reviewer
Google Scholar
Contact Name
Aji Prasetya Wibawa
Contact Email
keds.journal@um.ac.id
Phone
+62818539333
Journal Mail Official
keds.journal@um.ac.id
Editorial Address
Semarang St. No. 5, Malang, East Java, 65145, Indonesia
Location
Kota malang,
Jawa timur
INDONESIA
Knowledge Engineering and Data Science
Published by Universitas Negeri Malang
ISSN : -     EISSN : 25974637     DOI : http://dx.doi.org/2597-4637
Core Subject : Science, Engineering,
Computer Science & IT Engineering
The journal welcomes experimental and theoretical findings on data science and knowledge engineering along with their applications to real-life situations.
Arjuna Subject : Ilmu Komputer (semua kategori) - Sistem Informasi
Articles 117 Documents
 7   8   9   10   11 
From Data to Insight: A Machine Learning Approach in Classifying Dairy Cow Productivity Level and Identifying Important Influencing Variables Fauzi, Fatkhurokhman; Fauzan, Achmad; Widiyanto, Rhendy K P; Notodiputro, Khairil Anwar; Sartono, Bagus
Knowledge Engineering and Data Science
Publisher : citeus

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Identifying influential predictor variables is crucial for enhancing model interpretability in supervised classification. This study applies Permutation Variable Importance (PVI), a model-agnostic approach, to evaluate variable relevance after model fitting. Using data from the 2024 Indonesia Dairy Cow Productivity Survey, this research investigates five classification techniques: (1) Support Vector Machine (SVM), (2) Neural Network (NN), (3) k-Nearest Neighbors (kNN), (4) Naïve Bayes Classifier (NB), and (5) Logistic Regression (LR), to identify which method(s) yield the best performance based on evaluation metrics such as accuracy, sensitivity, and specificity. PVI is employed to identify the most influential predictor variables within the best-performing classification method. The novelty of this study lies in integrating model-agnostic interpretability with multiple supervised classifiers to generate transparent, data-driven insights into dairy productivity determinants. Results indicate that the top-performing methods, SVM and NN, achieved predictive accuracies ranging from 70% to 89%. Specifically, the SVM model achieved an accuracy of 0.799, a precision of 0.845, and an F1-score of 0.795, while the NN model obtained an accuracy of 0.786, a precision of 0.806, and an F1-score of 0.791. A permutational multivariate analysis of variance (PERMANOVA) on evaluation metrics revealed no statistically significant difference between the two methods. By applying PVI, nine key variables were consistently highlighted by both models as significant predictors for classifying dairy cow productivity levels (e.g., high vs. low yield) in Indonesia. These variables include farm altitude, the numbers of dairy heifers, lactating cows, and dry cows, the average duration of lactation and dry periods per cow annually, the daily amounts of forage, concentrate, and agricultural by-product feed provided per cow. These findings not only enhance model interpretability but also offer practical guidance for farm-level decision-making, the development of data-driven decision support systems, and the design of targeted policy interventions to improve dairy productivity in Indonesia, demonstrating the real-world applicability of machine-learning-based insights to strengthen dairy farm performance.
Recurrent Session Approach to Generative Association Rule based Recommendation Armanda, Tubagus Arief; Wardhani, Ire Puspa; Akhriza, Tubagus M.; Admira, Tubagus M. Adrie
Knowledge Engineering and Data Science
Publisher : citeus

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

This article introduces a generative association rule (AR)-based recommendation system (RS) using a recurrent neural network approach implemented when a user searches for an item in a browsing session. It is proposed to overcome the limitations of the traditional AR-based RS which implements query-based sessions that are not adaptive to input series, thus failing to generate recommendations. The dataset used is accurate retail transaction data from online stores in Europe. The contribution of the proposed method is a next-item prediction model using LSTM, but what is trained to develop the model is an associative rule string, not a string of items in a purchase transaction. The proposed model predicts the next item generatively, while the traditional method discriminatively. As a result, for an array of items that the user has viewed in a browsing session, the model can always recommend the following items when traditional methods cannot. In addition, the results of user-centered validation of several metrics show that although the level of accuracy (similarity) of recommended products and products seen by users is only 20%, other metrics reach above 70%, such as novelty, diversity, attractiveness and enjoyability.
Time Series Forecasting with LSTM: an extensive content analysis Pranolo, Andri; Zhou, Xiaofeng; Mao, Yingchi
Knowledge Engineering and Data Science
Publisher : citeus

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

This paper presents a comprehensive bibliometric and content review of the trend, architecture, and application of long short-term memory (LSTM) models for time series forecasting. The study aims to provide insights into the overall statistics and distribution of papers focused on LSTM for forecasting. Additionally, the research questions address the most highly cited papers based on LSTM approaches in forecasting, the most productive journals in this field, and identifying trends, gaps, summary tasks and their performance, datasets availability, and future research directions for LSTM in forecasting. This paper is a comprehensive review of LSTM for forecasting from 2017 to 2023 and identifies emerging trends, potential research gaps, and future directions for LSTM in forecasting. These findings contribute to a deeper understanding of the current state of LSTM-based forecasting research and provide valuable insights for researchers and practitioners in the field. This bibliometric and content review sheds light on the landscape of LSTM for time series forecasting, highlighting the most cited papers and productive journals and outlining potential areas for future exploration and development of LSTM models in forecasting.
Stacked LSTM-GRU Long-Term Forecasting Model for Indonesian Islamic Banks Sujatna, Yayat; Karno, Adhitio Satyo Bayangkari; Hastomo, Widi; Yuningsih, Nia; Arif, Dody; Handayani, Sri Setya; Kardian, Aqwam Rosadi; Wardhani, Ire Puspa; Rere, L.M Rasdi
Knowledge Engineering and Data Science
Publisher : citeus

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

The development of the Islamic banking industry in Indonesia has become a significant concern in recent years, with rapid growth in the number of banks operating based on Sharia principles. To face emerging challenges and opportunities, a deep understanding of the long-term financial behavior of Islamic banks is becoming increasingly important. This study aims to predict the share price of PT Bank Syariah Indonesia Tbk, over 28 days using the LSTM-GRU stack. The observation stage includes importing the dataset, data separation, model variations, the training process, output, and evaluation. Observations were conducted using 10 model variations from 4 stacks of LSTM and GRU. Each model performs the training process in four epochs (200, 500, 750, and 1000). The results of observations in this study show that long-term predictions (28 days ahead) using four stacks of LSTM-GRU and daily training accumulation techniques produce better accuracy than the general method (using multiple outputs). From the observations we have made for predictions for the next 28 days, the model with the LGLG stack arrangement (LSTM-GRU-LSTM-GRU) produces the best accuracy at epoch 750 with an MSE LSTM-GRU 63.43762863. This study will undoubtedly continue in order to achieve even better precision, either by utilizing a new design or by further improving the technology we are now employing.
Real-time Deep Learning Detection of Toraja Carving Motifs using YOLO11m for Cultural Heritage Preservation Herman, Herman; Mufti, Farid Wajdi; Manga, Abdul Rachman; Nasir, Haidawati
Knowledge Engineering and Data Science
Publisher : citeus

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Toraja carvings are an important part of Indonesia’s cultural heritage, rich in symbolic, aesthetic, and philosophical meaning. However, the identification and preservation of carving motifs still rely on subjective, time-consuming manual processes, limiting scalability and inconsistent knowledge transmission. From a Knowledge Engineering and Cognitive Data Science perspective, this challenge highlights the need for mechanisms that can transform visual cultural artifacts into structured, machine-interpretable knowledge. This study investigates the use of the YOLO11m model as a data-driven approach for modeling cultural knowledge through automated detection of three Toraja carving motifs: pa_tedong, pa_kapu_baka, and pa_manu_londongan using original images collected directly from traditional Tongkonan houses. A total of 222 high-resolution images were manually annotated and preprocessed through resizing to 640×640 pixels and auto-orientation for input standardization, then divided into 70% training, 20% validation, and 10% testing sets. The model was trained on an NVIDIA A100 GPU using the Ultralytics framework in Google Colab and evaluated using precision, recall, F1-score, and mean Average Precision (mAP). Experimental results show that YOLO11m achieved an mAP@0.5 of 96.5%, demonstrating robust performance in capturing complex visual and semantic patterns despite challenges such as motif similarity and data imbalance. Beyond detection accuracy, the findings indicate that deep learning–based object detection can support the systematic documentation, interpretation, and reuse of cultural knowledge, contributing to scalable digital preservation of traditional cultural artifacts. Future work will explore larger datasets, optimized hyperparameters, and advanced detection models to further enhance the robustness of cultural knowledge representations further.
Comparative Analysis of Deep Learning Algorithms for Predicting ENSO Based on Non-sequential Sampling Procedure Algorithms Faulina, Ria; Hasanah, Siti Hadijah; Nuramaliyah, Nuramaliyah; Fitriana, Ika Nur Laily
Knowledge Engineering and Data Science
Publisher : citeus

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

The El Niño–Southern Oscillation (ENSO) is a major climate phenomenon that significantly influences global weather patterns, particularly rainfall and temperature variability across different regions. Accurate ENSO forecasting is therefore essential to support disaster risk mitigation and strategic decision-making in climate-sensitive sectors such as agriculture, fisheries, and water resource management. This study investigates the performance of deep learning approaches for ENSO prediction using a non-sequential sampling procedure on historical climate data. Three models are comparatively evaluated: Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and a hybrid CNN–LSTM architecture. The results demonstrate that the hybrid CNN–LSTM model outperforms the standalone CNN and LSTM models in predictive accuracy and robustness. Specifically, the proposed model achieved the lowest Mean Absolute Error (MAE) of 13.97 and Root Mean Square Error (RMSE) of 15.76 across multiple test samples. These findings indicate that the integration of convolution-based feature extraction and sequential memory learning effectively captures complex ENSO temporal patterns. The proposed approach offers a reliable computational framework for climate forecasting and may contribute to improved anticipatory planning in climate-sensitive decision-making contexts.
Cognitive EEG Differentiation with Hypnosis-Based Noise Reduction and K-Harmonic Means for Personalized Brainwave Modeling Azhari, Ahmad; Saputra, Dimas Chaerul Ekty
Knowledge Engineering and Data Science
Publisher : citeus

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

This study investigates the integration of hypnosis-based noise reduction and K-Harmonic Means (KHM) clustering for personalized brainwave modeling using Electroencephalography (EEG) data. EEG signals were collected from 100 participants using a Neurosky Mindset sensor at the FP1 (prefrontal) location, with each subject performing nine standardized cognitive tasks such as breathing, memory recall, and mathematical problem-solving. Hypnosis was applied not as a filtering method but as a behavioral protocol to standardize subject conditions and minimize physiological and environmental noise. The EEG signals were sampled at 128 Hz and analyzed using KHM clustering with K=4K = 4K=4, resulting in a Silhouette Score of 0.9515, which demonstrates strong cluster separation and robustness against noise. Compared with baseline approaches such as K-Means and Fuzzy C-Means, KHM achieved higher stability and consistency in differentiating cognitive tasks. This performance highlights the advantage of harmonic averaging in mitigating the influence of outliers during clustering. The findings suggest that hypnosis can meaningfully enhance EEG signal quality, thereby improving downstream cognitive state differentiation. Overall, this research contributes to advancing EEG-based cognitive analysis and personalized brainwave modeling, with potential applications in brain–computer interfaces, cognitive diagnostics, and neurofeedback systems. The integration of behavioral noise control (hypnosis) with advanced clustering methods presents a novel hybrid framework for improving the reliability of EEG-based cognitive state identification.
Comparative Performance of VGG16 and EfficientNetB0-Based Transfer Learning for Brain Tumor Classification Azis, Huzain; Jalil, Rizqi Ananda; Manga', Abdul Rachman
Knowledge Engineering and Data Science
Publisher : citeus

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

The classification of brain tumors using Magnetic Resonance Imaging (MRI) images is essential for early diagnosis but remains challenging due to tumor diversity. This study evaluates the effectiveness of two distinct architectural approaches for feature extraction: VGG16, representing a classic sequential design, and EfficientNetB0, a modern architecture optimized for parameter efficiency through compound scaling. Using a dataset of 2,870 MRI images categorized into four classes, we implemented a static transfer learning strategy by freezing all pre-trained ImageNet weights to act as fixed feature extractors. Features were extracted from specific layers, the final pooling layer for VGG16 and the Global Average Pooling (GAP) layer for EfficientNetB0. To optimize the discriminative power of these static features, this study implements an ensemble-fusion framework, specifically Voting and Stacking classifiers, to integrate the strengths of diverse base learners (SVM, Random Forest, and XGBoost) in identifying complex brain tumor patterns. Results demonstrate that VGG16-based models consistently and significantly outperform EfficientNetB0-based models across all evaluation metrics. The VGG16 + Random Forest combination achieved a peak accuracy of 73.9%, whereas EfficientNetB0-based models struggled, with peak accuracies reaching only 36.1%. Statistical analysis confirms that VGG16’s uniform  convolutional layers provide more stable and discriminative feature representations for medical textures compared to the mobile-optimized blocks of EfficientNetB0 when weights are not fine-tuned. This research highlights that, for static feature extraction in resource-constrained environments, classic architectural stability offers superior reliability compared to modern parameter efficiency.
Comparison of Machine Learning Algorithms for Species Family Classification using DNA Barcode Riza, Lala Septem; Rahman, M Ammar Fadhlur; Prasetyo, Yudi; Zain, Muhammad Iqbal; Siregar, Herbert; Hidayat, Topik; Abu Samah, Khyrina Airin Fariza; Rosyda, Miftahurrahma
Knowledge Engineering and Data Science
Publisher : citeus

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Classifying plant species within the Liliaceae and Amaryllidaceae families presents inherent challenges due to the complex genetic diversity and overlapping morphological traits among species. This study explores the difficulties in accurate classification by comparing 11 supervised learning algorithms applied to DNA barcode data, aiming to enhance the precision of species family classification in these taxonomically intricate plant families. The ribulose-1,5-bisphosphate carboxylase-oxygenase large sub-unit (rbcL) gene, selected as a DNA barcode locus for plants, is used to represent species within the Amaryllidaceae and Liliaceae families. The experimental results demonstrate that nearly all tested models achieve accurate species classification into the appropriate families, with an accuracy rate exceeding 97%, except for the Naïve Bayes model. Regarding computational time, the Random Forest model requires significantly more time for training than other models. Regarding memory usage, the Least Squares Support Vector Machine with a polynomial kernel, and Regularized Logistic Regression consume more memory than other models. These machine learning models exhibit strong concordance with NCBI's classifications when predicting families using the test dataset, effectively categorizing species into the Amaryllidaceae and Liliaceae families.
Deep Learning for Multi-Structured Javanese Gamelan Note Generator Kurniawati, Arik; Yuniarno, Eko Mulyanto; Suprapto, Yoyon Kusnendar
Knowledge Engineering and Data Science
Publisher : citeus

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Javanese gamelan, a traditional Indonesian musical style, has several song structures called gendhing. Gendhing (songs) are written in conventional notation and require gamelan musicians to recognize patterns in the structure of each song. Usually, previous research on gendhing focuses on artistic and ethnomusicological perspectives, but this study is to explore the correlation between gendhing as traditional music in Indonesia and deep learning technology that replaces the task of gamelan composers. This research proposes CNN-LSTM to generate notation of ricikan struktural instruments as an accompaniment to Javanese gamelan music compositions based on balungan notation, rhythm, song structure, and gatra information. This proposed method (CNN-LSTM) is compared with LSTM and CNN. The musical data in this study is represented using numerical notation for the main melody in balungan notation. The experimental results showed that the CNN-LSTM model showed better performance compared to the LSTM and CNN models, with accuracy values of 91.9%, 91.5%, and 91.2% for CNN-LSTM, LSTM, and CNN, respectively. And the value of note distance for the Sampak song structure is 4 for the CNN-LSTM model, 8 for the LSTM model, and 12 for the CNN model. The smaller the note distance, the closer it is to the original notation provided by the gamelan composer. This study provides relevance for novice gamelan musicians who are interested in learning karawitan, especially in understanding ricikan struktural music notation and gamelan art in composing musical compositions of a song.

Page 9 of 12 | Total Record : 117

 7   8   9   10   11 

Filter by Year

2018 2025


Reset
Filter By Issues
All Issue