Publication Search

 Cryptocurrency price prediction, particularly for highly volatile assets like Solana (SOL), is a crucial challenge in time series data analysis in digital finance. This study aims to compare the performance of the XGBoost machine learning algorithm with the Temporal Fusion Transformer (TFT) deep learning model in predicting Solana's daily closing price. The dataset used consists of historical Solana price data and network fundamentals data in the form of Total Value Locked (TVL). The research process includes data preprocessing, dividing training and test data, model training, and evaluation using the Root Mean Squared Error (RMSE) metric. The results show that using the same-day price feature has the potential to cause target leakage, resulting in invalid prediction accuracy. In testing using pure historical data without data leakage, the XGBoost model performed better than TFT with an RMSE of 4.27, while TFT produced an RMSE of 18.59. Furthermore, the integration of network fundamentals data in the form of TVL did not improve prediction accuracy and even caused a decrease in performance for the XGBoost model with an RMSE of 7.10. The results of this study show that the use of historical price action features is more effective than fundamental network indicators for short-term daily Solana price predictions.

Accurate classification of rainfall intensity patterns is important for early warning systems, hydrometeorological risk assessment, and water resource management. Surface rain gauges have limited spatial coverage, so this study uses NOAA NEXRAD Level II radar data from the KTLX station in 2023. K-Means clustering was applied to identify rainfall intensity patterns from 30 randomly selected days, with scans stratified into four daily time intervals. Seven features were extracted from each radar sweep, including reflectivity statistics, convective and stratiform ratios, and rainfall coverage. The data were normalized and balanced before clustering. The optimal cluster count was determined through a combined evaluation of the Elbow Method, Silhouette Score, and Davies-Bouldin Index, yielding K=5 as the most representative configuration. Evaluation results demonstrated a Silhouette Score of 0.3871 and a Davies-Bouldin Index of 0.8599, indicating moderate cluster cohesion that reflects the inherent overlapping nature of rainfall intensity transitions in radar reflectivity data. The clusters represent rainfall regimes from non-precipitating conditions to intense convective events. These results support the use of K-Means for automated rainfall pattern recognition and flood forecasting applications. 

The architecture of consumer decision-making has completely changed due to the quick development of recommendation systems based on artificial intelligence (AI). The majority of earlier studies saw algorithms as instruments for forecasting and maximizing preexisting preferences. This study, however, makes a different claim: algorithmic curation actively shapes preferences rather than just reflecting them. This study creates and evaluates a structural model that examines the impact of algorithmic curation intensity on perceived search autonomy, identity resonance, affective evaluation, and the development of initial preferences. The model is based on identity-based consumption theory and the literature on human-AI interaction. The study's findings, which are based on survey data from Generation Z consumers and Structural Equation Modeling (SEM) analysis, demonstrate a contradictory dynamic: algorithmic curation improves identity resonance and directly influences initial preferences while simultaneously decreasing feelings of autonomy. The primary mediating mechanism that links algorithmic exposure to emotional assessment and preference creation is identified as identity resonance. In addition to introducing the concept of algorithmic consumer formation as a new conceptual framework for comprehending consumer behavior in the AI-based digital era, our findings expand the notion of bounded rationality toward algorithmically bounded agency.

This study aims to analyze the fluctuations in chili production in Kabanjahe District, Karo Regency, which affect market price instability and uncertain supply. One approach applied in this study is the Single Exponential Smoothing (SES) method to forecast chili production. SES was chosen for its simplicity, ease of implementation, and its ability to generate accurate predictions even when the data lacks significant seasonal patterns. The data used is secondary data on chili production obtained from official publications by the Karo Regency BPS for the period of 2020–2024. The analysis results show that a smoothing parameter (α) of 0.8 produced the lowest Mean Absolute Percentage Error (MAPE) of 3.08%. These findings indicate that applying a higher α makes the model more responsive to recent data changes, thus yielding more accurate forecasts. This study demonstrates the effectiveness of the SES method in forecasting chili production in areas with significant seasonal fluctuations.

The retention of customers in the retail banking sector is a critical economic imperative; however, predictive modeling is frequently hindered by severe class imbalance and the “Black Box” nature of complex algorithms. This study proposes a Heterogeneous Stacking Ensemble framework integrating XGBoost, CatBoost, and Random Forest base learners with a Logistic Regression meta-learner to forecast customer attrition. To overcome the pervasive “Majority Class Bias,” we introduce a “Dual-Imbalance Defense” that synergizes the Synthetic Minority Over-sampling Technique (SMOTE) with algorithmic cost-sensitive penalization. Furthermore, moving beyond standard accuracy metrics, the framework mathematically derives a dynamic classification threshold to guarantee a strict 0.90 recall rate, actively optimizing the capture of at-risk capital. Model opacity is addressed through the integration of a SHapley Additive exPlanations (SHAP) TreeExplainer. This cooperative game theory approach provides localized, patient-level “Reason Codes” for regulatory compliance and reveals global systemic vulnerabilities, including non-linear drivers such as the “Product Paradox.” Achieving a 0.90 recall rate and an AUC of 0.8654, this framework provides a statistically robust and operationally transparent tool for targeted customer retention.

Attention mechanisms have been widely incorporated into recurrent neural network architectures for financial time series forecasting, with most prior work reporting improvements in price-level error metrics. This study revisits that claim through a controlled empirical comparison of four deep learning architectures on nearly two decades of Telkom Indonesia (TLKM) closing price data from the Indonesia Stock Exchange (IDX). The models evaluated are a three-layer Gated Recurrent Unit (GRU) baseline, a comparable Long Short-Term Memory (LSTM) network, a Bahdanau end-attention GRU (Attn-GRU-V2), and a multi-head self-attention GRU hybrid (Attn-GRU-V3). Each architecture is trained over 30 independent runs with distinct random seeds, and performance is reported as 95% confidence intervals derived from the t-distribution. Statistical comparisons employ the Wilcoxon signed-rank test, a nonparametric paired test appropriate given the confirmed non-normality of residuals. The main finding is a consistent trade-off: the plain GRU achieves the lowest RMSE (94.02 ± 1.22 IDR) across all 30 runs, while Attn-GRU-V2 achieves the highest directional accuracy (45.91 ± 0.09%), surpassing GRU in every independent run. Bahdanau attention weights are nearly uniform across the 30-day lookback window (coefficient of variation: 3.21%), indicating that the mechanism cannot identify selectively informative timesteps in this univariate price series. This finding is consistent with the weak-form Efficient Market Hypothesis for the Indonesian market. An ablation study reveals that a 20-day lookback window maximizes directional accuracy (47.72 ± 0.21%) for the Attn-GRU-V2 model. These results suggest that Bahdanau end-attention consistently and significantly improves directional accuracy relative to a plain GRU baseline, providing an architecturally attributable advantage for direction-based applications, even when absolute price-level error is not reduced. The directional accuracy values remaining below 50% across all models are consistent with a weak-form efficiency characterization of the Indonesian market.

Ineffective spare part inventory planning may lead to supply delays and reduced compliance with lead time supply key performance indicators (KPIs). This study aims to implement the Collaborative Planning, Forecasting, and Replenishment (CPFR) method in spare part inventory planning at PT XYZ and to compare lead time supply performance before and after the implementation of the CPFR method. This research utilizes spare part usage data from January to June 2025, focusing on fast-moving spare parts. Demand forecasting is conducted using an error forecasting approach with the moving average method. Forecast accuracy is evaluated using the Mean Absolute Deviation (MAD) and Mean Absolute Percentage Error (MAPE). Furthermore, inventory planning is carried out through the calculation of safety stock and reorder point (ROP) as the basis for determining replenishment decisions. The results indicate that the simulated implementation of the CPFR method provides a more structured and anticipative inventory planning process. The comparison of performance before and after the application of CPFR shows an improvement in lead time supply compliance with the established KPIs. Therefore, the CPFR method has the potential to support improved spare part inventory planning performance at PT XYZ.

The idea of functions in mathematics and how they are used to build different mathematical models are methodically examined in this publication. Functions are basic mathematical constructs that show relationships between two or more variables in explicit equations, tables, or graphs. The fundamental building blocks of mathematics are functions, which enable the representation of variable interdependencies in a variety of formats, including formal mathematical expressions, data tables, and graphs. The classification of function types, such as linear, quadratic, and exponential, and their corresponding uses in the domains of physics, economics, and epidemiology are the main topics of this study, which takes a descriptive and exploratory approach.This article illustrates how knowledge of functions greatly aids processes through a review of the literature and an examination of secondary sources from current textbooks and academic publications. of judgment, forecasting, and analysis. In both academic and professional contexts, mathematical modeling based on functions has demonstrated efficacy in accurately and efficiently representing real-world occurrences. Thus, the significance of incorporating functional thinking into STEM education and multidisciplinary practice is emphasized in this essay.

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.

Micro, Small, and Medium Enterprises (MSMEs) play a vital role in driving economic growth; however, their production activities frequently face uncertainty in achieving predetermined targets. Such uncertainty arises from fluctuating market demand, delays in raw material supply, labor limitations, variations in processing time, and other technical constraints. Conventional deterministic production planning methods often fail to capture these real-world risks and variations, leading to less accurate and suboptimal decisions. Therefore, a more adaptive analytical approach that incorporates probability and uncertainty is required. This study aims to analyze the probability of achieving MSME production targets using the Monte Carlo Simulation method. This method models random production conditions by generating data based on probability distributions derived from historical records. Simulations are repeated through numerous iterations to estimate possible variations in production output and measure the likelihood of meeting targets. The results indicate that Monte Carlo simulation provides more realistic and comprehensive production forecasts compared to traditional planning approaches. By understanding both the probability of success and potential risks, MSMEs can design adaptive strategies, optimize resource allocation, manage inventory more effectively, and improve overall production planning accuracy to ensure long-term business sustainability in a dynamic environment.

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.

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.

Rainfall is a crucial factor in the stability of the Earth's ecosystem and has a significant impact on agriculture, forestry, energy, and water management. However, increasingly unstable climate change makes rainfall patterns difficult to predict accurately using traditional methods. The city of Medan, the capital of North Sumatra Province, has a tropical rainforest climate with an average annual rainfall of approximately ±2200 mm and an average temperature of 27°C. Significant weather fluctuations in this area can trigger flooding when rainfall increases and cause water shortages when rainfall decreases (BMKG, 2021). Therefore, a prediction approach that can manage non-linear and dynamic data is needed. Artificial Neural Networks (ANN) are one of the reliable machine learning methods for detecting data patterns. By using the backpropagation algorithm, the model can gradually reduce prediction errors, making it widely used in weather forecasting applications. In this regard, this study uses ANN with the backpropagation method to forecast monthly rainfall in Medan City by utilizing data from 2022–2024 as training and testing data.

In the rapidly evolving digital economy, the ability to anticipate transaction surges is a strategic asset for marketplace platforms to maintain operational efficiency. This research aims to build an accurate daily transaction volume forecasting system thru the implementation of an Extract, Transform, and Load (ETL) pipeline and Autoregressive Integrated Moving Average (ARIMA) predictive modeling. The dataset used is sourced from dataset_olshop.csv, which includes transaction history for the entire year of 2025. The ETL stage focused on data cleaning and handling missing values, while time series analysis began with the Augmented Dickey-Fuller (ADF) stationarity test, which yielded a significant p-value of 0.000006. The parameter model was optimized using the auto_arima algorithm, which determined the ARIMA(2,0,0) configuration as the best model. The evaluation results of the model show fairly stable performance with a Root Mean Squared Error (RMSE) value of 2.002 and a Mean Absolute Error (MAE) of 1.704 on the test data. Research findings reveal a consistently higher purchasing pattern during the mid-month and end-of-month periods, with an average of 5.52 daily transactions, compared to the beginning of the month, which saw 5.48 transactions. The 30-day forecast results provide valuable insights for online store managers to proactively adjust inventory and logistics workforce allocation strategies. This research concludes that integrating data engineering techniques and statistical analysis can provide predictive solutions for the dynamics of the digital market.

This study examines the development of a Crude Palm Oil (CPO) price forecasting model using an artificial neural network algorithm, specifically the backpropagation algorithm. As one of Indonesia’s main export commodities, CPO has a significant economic impact and influences the income of oil palm farmers. The CPO price data used in this study were obtained from CIF Rotterdam, covering the period from January 2019 to December 2023. The research methodology consists of several stages, including data collection, preprocessing, model design, and model implementation using Python programming. The training results of the backpropagation algorithm show an error value of 0.537829578 after 1,000 epochs, while the evaluation using Mean Squared Error (MSE) indicates an MSE of 0.022709 during the training process and 0.017604 during the testing process. The model also produces CPO price predictions for the next three months, namely 932.578 for the first month, 949.568 for the second month, and 774.855 for the third month. These findings indicate that the developed model is capable of predicting future CPO prices with adequate accuracy, which can assist companies in making better financial decisions and managing risks associated with CPO price fluctuations.

The dynamic fluctuations in stock prices present a major challenge for investors in making informed decisions. To anticipate such uncertainties, forecasting methods that can provide accurate predictions are required. This study compares two time series forecasting methods Autoregressive Integrated Moving Average (ARIMA) and Double Exponential Smoothing (Holt) in predicting the stock prices of PT Telkom Indonesia (TLKM). The dataset consists of monthly closing prices from January 2018 to December 2023. The performance of each model is evaluated using three error metrics: Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE). The results show that the ARIMA(1,1,1) model yields higher predictive accuracy than the Holt method, with MAE of 787.71, MSE of 771,844.2, and RMSE of 878.55. In contrast, the Holt method records a MAE of 837.19, MSE of 878,393.4, and RMSE of 937.23. These findings confirm that ARIMA is superior in capturing the complex patterns of stock price movements and is more effective in volatile market conditions such as the stock exchange.

The relatively high volatility of Robusta coffee prices creates uncertainty for farmers, business actors, and policymakers in making economic decisions. This study aims to analyze the price movement patterns of Robusta coffee, determine the most appropriate Autoregressive Integrated Moving Average (ARIMA) model, and conduct short- to medium-term price forecasting for Robusta coffee. The data used consist of monthly Robusta coffee price data from January 2023 to September 2025, sourced from the World Bank Commodity Price Data. The analytical method employed is ARIMA using EViews software, beginning with stationarity testing using the Augmented Dickey-Fuller (ADF) test, model identification through ACF and PACF, parameter estimation, and residual diagnostic testing. The results show that Robusta coffee price data are non-stationary at the level but become stationary at the first difference, indicating integration of order one I(1). Based on model identification and diagnostic testing, the ARIMA (0,1,0) model is found to be the most appropriate and satisfies the white noise assumption. Forecasting results indicate that Robusta coffee prices are projected to remain relatively stable with a moderate upward trend through December 2026. These findings are expected to serve as a reference for decision-making by farmers, business actors, and the government in responding to Robusta coffee price dynamics.

This study analyzes and designs a dual-role web-based ordering information system to optimize order management at Sunrise Bakery. This SME currently faces inefficiencies due to manual recording. The system, developed using the SDLC Waterfall method with PHP and MySQL, serves two main actors: customers, who can order online, browse catalogs, track orders, and pay digitally; and administrators (admin, cashier, owner), who manage products, update stock, input in-store orders, generate daily/monthly sales reports, and manage user access. Black Box Testing confirms all core functions work correctly. The system successfully addresses manual process shortcomings by improving data accuracy and providing real-time monitoring for both customers and management. It offers a comprehensive digital solution to enhance operational efficiency and service quality. Limitations include the lack of integrated digital payment gateways and external messaging. Future development should incorporate payment gateways (e.g., OVO, GoPay), WhatsApp notifications, a mobile application, and predictive analytics for sales and stock forecasting.

The rapid development of information technology in today's digital era has significantly impacted organizational performance, particularly in data management and resource planning. One organization that heavily relies on accurate data availability is the Indonesian Red Cross (PMI), especially its Blood Donor Unit (UDD). UDD PMI of Asahan Regency faces challenges in determining monthly blood donor targets to maintain stable blood stock. A shortage of blood supply can be fatal for patients requiring transfusions. Therefore, a system is needed to forecast the number of blood donors, allowing for more accurate decision-making. This study utilizes the Weighted Moving Average (WMA) method to predict the number of blood donors for the following month based on historical data from March 2024 to March 2025. The WMA method is chosen for its ability to assign greater weight to recent data, making the forecast more relevant and accurate. The results of this research are expected to assist UDD PMI Asahan Regency in anticipating blood needs and maintaining optimal stock availability.

The instability of the climate is becoming increasingly prominent across Southeast Asia, creating uncertainty in agricultural systems that are highly dependent on seasonal weather patterns. Indonesia, where rice remains the primary staple food, is particularly vulnerable to the effects of rising temperatures and rainfall deficits. This study applies the Seasonal Autoregressive Integrated Moving Average (SARIMA) model to predict rice production while incorporating indicators of extreme climate anomalies. Using publicly available datasets, including FAOSTAT production statistics, NOAA rainfall and temperature anomalies, and climate indices from the World Bank, this model was developed following the Box-Jenkins procedure. Among the configurations tested, the SARIMA model (1,1,1)(0,1,1)₁₂ showed the strongest performance, reflected in a MAPE of 4.62% and low RMSE values. The model indicates that significant El Niño events can reduce annual rice production by 3–7%, while wetter La Niña conditions may support production recovery. These findings highlight the importance of integrating climate-sensitive data into agricultural forecasting. The model presented here could support early warning systems, adaptive farming strategies, and long-term food security planning in Indonesia.