Publication Search

Predictive maintenance (PdM) plays a crucial role in modern industrial systems by minimizing downtime, reducing maintenance costs, and optimizing asset performance. However, many predictive models operate as “black box” systems, limiting transparency and making it difficult for operators to interpret their outputs. This study aims to integrate Explainable Artificial Intelligence (XAI) techniques with Remaining Useful Life (RUL) prediction models to improve both accuracy and interpretability. Various machine learning and deep learning approaches, including Support Vector Machines (SVM), Random Forest (RF), XGBoost, Long Short-Term Memory (LSTM), and Convolutional Neural Networks (CNN), are employed to predict RUL using real-time sensor data from rotating machinery. XAI methods such as SHAP, LIME, and attention mechanisms are applied to provide human-understandable explanations of model predictions. The models are evaluated based on accuracy, Root Mean Square Error (RMSE), and interpretability scores. The results show that XAI-enhanced models outperform traditional approaches in predictive performance while offering greater transparency. These explanations help maintenance engineers better understand the factors influencing predictions, thereby improving decision-making and trust in the system. Nevertheless, the integration of XAI introduces additional computational complexity, which may pose challenges for large-scale industrial implementation. Overall, this study highlights the potential of combining XAI with RUL prediction to develop more reliable, transparent, and effective predictive maintenance solutions.

The proliferation of online gambling promotional comments on Indonesian social media has become a serious issue requiring fast and accurate automated handling. This study aims to implement a Hybrid Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM) method to classify online gambling comments and compare its performance with standalone RNN and LSTM models. The research utilized a dataset of 10,230 comments subjected to comprehensive preprocessing stages, including the normalization of non-standard language using a slang dictionary. Testing was conducted across three data-splitting scenarios: 90:10, 80:20, and 70:30. Experimental results demonstrate that the standalone LSTM model achieved the highest average accuracy of 97.45%. However, the Hybrid RNN–LSTM model showed significant superiority in terms of performance stability, yielding the lowest standard deviation (0.0027) and the smallest Coefficient of Variation (0.28%) across all scenarios. These findings indicate that while the LSTM architecture is highly effective at capturing short-text context, the Hybrid approach provides better robustness against fluctuations in data proportions, making it highly relevant for implementation as an automated detection system on social media.

The increasing sedentary lifestyle in the digital era has the potential to cause various health problems due to lack of physical activity. One approach that can be taken to encourage physical activity is through the use of digital games with body movement-based control mechanisms. This study aims to develop a body gesture-based game character control system using a hybrid Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) model. CNN is used to extract spatial features from each video frame, while LSTM serves to model the temporal relationship between frames so that movement patterns can be recognized sequentially. The research method used refers to the Machine Learning Lifecycle stages, starting from data collection, preprocessing, model development, to implementation in the endless runner game genre. Testing results show that the CNN–LSTM model is capable of classifying body gestures and generating outputs that can be used as commands to control game characters. The implementation of this system enables more natural and interactive game interactions without conventional input devices, and has the potential to encourage players to lead a more active lifestyle.

International student enrollment is a critical driver of financial sustainability for Higher Education Institutions (HEIs). While advanced forecasting is standard in the corporate sector, its application in educational planning remains limited. This study addresses this gap by comparing the predictive performance of ARIMA, Facebook Prophet, and Long Short-Term Memory (LSTM) models. Using a publicly available annual dataset from a US-based institution (2000–2022), the analysis employed a strategic partition training on 2000–2017 and testing on 2018–2019 to validate models on stable, pre-pandemic data. Empirical results revealed that the statistical ARIMA (2,1,0) model demonstrated superior accuracy, achieving a Mean Absolute Percentage Error (MAPE) of 1.26%. Conversely, Prophet (11.81%) and LSTM (13.84%) struggled with the limited sample size, failing to generalize effectively compared to the linear approach. The findings suggest that for annual enrollment trends, parsimonious statistical models outperform complex deep learning architectures, providing administrators with a robust, accessible framework for data-driven strategic decision-making.

Indonesia, as a country with the highest seismicity in the world, requires an accurate earthquake prediction system through the use of the BMKG earthquake catalogue. This research aims to implement ETL-based data pipeline engineering to process 92,887 earthquake catalog entries for the 2008-2023 period into ready-to-use daily time series for the LSTM seismicity forecasting model. The ETL process includes raw data extraction, cleaning of 97% missing values columns on focal mechanism parameters, datetime conversion, daily resampling producing 5,200 entries with earthquake count, total magnitude, and average magnitude features, as well as Min-Max Scaler normalization for LSTM compatibility. The dataset was processed using Google Colab with a stacked LSTM architecture of two layers of 50 and 25 units, dropout 0.2, Adam optimizer, and a sequence window of 30 days to predict the daily earthquake count. The model trained for 100 epochs shows the ability to capture stable seismic activity trends with a consistent decrease in MSE loss, although it shows deviations in extreme spikes due to aftershock sequences. The ETL pipeline proved crucial in ensuring temporal consistency, 100% data completeness, and relevant physics representation, resulting in a reproducible end-to-end framework for disaster mitigation.

This study is motivated by the increasing need for accurate modeling and classification of one-dimensional signal data in intelligent systems. The rapid development of deep learning has led to the adoption of more adaptive and complex neural network architectures capable of capturing both temporal dependencies and local patterns in sequential data. This research aims to analyze and compare the performance of several deep learning models, namely Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and a hybrid Convolutional Neural Network–GRU (CNN–GRU) model for signal data classification. The research method employs a quantitative experimental approach involving data preprocessing, windowing, model training, and performance evaluation. Model performance is evaluated using accuracy, precision, recall, and F1-score metrics. The results indicate that the hybrid CNN–GRU model outperforms the other models, particularly in capturing local features and long-term temporal dependencies within signal data. These findings suggest that the integration of convolutional layers and recurrent mechanisms enhances feature representation and learning stability. This study is expected to contribute both theoretically and practically to the development of deep learning models for signal processing and time-series-based intelligent applications.

Stock price prediction remains a complex challenge due to the dynamic and non-linear nature of financial markets, especially for banking stocks like PT Bank Negara Indonesia (Persero) Tbk (BBNI). This study aims to optimize BBNI stock price forecasting by integrating an automated Extract, Transform, Load (ETL) pipeline with the Long Short-Term Memory (LSTM) algorithm within a data engineering framework. Historical data from 2019 to 2025 were processed through a structured ETL sequence—including data cleaning, feature engineering, and MinMaxScaler normalization—to ensure high data quality. The dataset was partitioned into 80% for model training and 20% for testing to ensure rigorous evaluation. The results demonstrate that the systematic ETL approach significantly enhances model stability and predictive accuracy compared to conventional methods. The LSTM model effectively captured long-term temporal dependencies, providing reliable trend forecasts with an impressive test accuracy, achieving a Root Mean Squared Error (RMSE) of 0.0354. This research underscores that integrating robust data engineering practices with deep learning is essential for building resilient financial decision-support systems.

The increasing complexity and scale of modern cybersecurity threats necessitate the development of advanced systems capable of efficiently detecting, analyzing, and mitigating incidents in real time. This paper proposes an automated framework for digital forensics and incident response that leverages big data analytics and real time network traffic profiling. The framework integrates cutting-edge technologies, including Apache Spark for real time data processing and Hadoop for scalable data storage, combined with machine learning models like LSTM and Autoencoders to detect anomalies and threats in network traffic. By automating the process of incident detection and response, this framework significantly reduces the time required to identify threats and improves the accuracy of forensic evidence correlation across heterogeneous network environments. The study highlights the advantages of using machine learning models and big data tools to address the limitations of traditional manual and semi-automated systems, which often struggle to keep pace with large-scale data generation. Testing results demonstrate that the proposed framework can handle large data volumes efficiently, providing real time, actionable insights with significantly reduced response times. Additionally, the framework improves forensic analysis by enabling the correlation of evidence from different devices and protocols, making it more effective than traditional methods in identifying the root cause of security incidents. However, challenges related to data heterogeneity, scalability, and system integration were encountered during testing. The proposed framework holds promise for significantly enhancing the efficiency and effectiveness of cybersecurity operations, with future work focusing on further integration of advanced AI techniques and machine learning models for dynamic and adaptive incident response.

User reviews on the Shopee e-commerce platform represent an important source of information for understanding consumer perceptions of products and services. Sentiment analysis is commonly applied to classify user opinions into positive, neutral, and negative sentiment categories based on textual data. This study aims to analyze the performance of the Long Short-Term Memory (LSTM) method in sentiment classification of Shopee user reviews. The dataset used in this study consists of Indonesian-language user reviews that have undergone preprocessing stages, including case folding, text cleaning, tokenization, and stopword removal. The LSTM model was trained using preprocessed text represented as word sequences. Model performance was evaluated using overall accuracy and class-wise classification results. The experimental results indicate that the LSTM method achieved an overall accuracy of 87.62%. In addition, the classification performance for the positive sentiment class reached 95.27%, the neutral class achieved 4.96%, and the negative class reached 74.26%. These results demonstrate that the LSTM method performs well in classifying sentiment in Shopee user reviews, particularly for positive sentiment. This study is expected to provide insights and references for the application of deep learning methods in sentiment analysis of Indonesian e-commerce review data.