Jurnal Aplikasi Statistika & Komputasi Statistik

Financial Literacy’s Impact on Interest in Sharia Investment: An Examination Using SEM-PLS

Sella Nofriska Sudrimo — 2026-06-30

Introduction/Main Objectives: This research explores the impact of financial literacy, including knowledge, confidence, and ability, on interest in Sharia investment. Background Problems: the relationship between financial literacy and interest in sharia investment. Novelty: Considering that each province has different community characteristics, especially in the Southwest Papua Province area which is included in the 3T areas (underdeveloped, frontier, and outermost), an analysis was carried out and using SEM-PLS for analysis of Ddta. Research Methods: Using the Structural Equation Model Partial Least Square (SEM-PLS) method with the response variable sharia investment interest and the financial literacy variable with the dimensions of knowledge, confidence and ability. Finding/Results: The results show that financial literacy has a positive and statistically significant effect on interest in Sharia investment. Although its explanatory power is limited (R² = 0.094), the findings indicate that financial literacy functions as an enabling factor rather than a sole determinant of Sharia investment interest. These results suggest that improving financial literacy alone is insufficient to substantially increase interest in Sharia investment. Therefore, policies aimed at promoting Sharia investment should integrate financial education with institutional trust-building, product accessibility, and socio-religious engagement, particularly in underdeveloped and frontier regions such as Southwest Papua.

Analysis of Factors Influencing Waste Generation in East Java

Syefa Ilmi Beandita Putri — 2026-06-30

Introduction/Main Objectives: Waste accumulation poses a serious threat to environmental sustainability and hinders the achievement of Sustainable Development Goal (SDG) 12 regarding responsible consumption and production patterns. Background Problems: East Java consistently ranks second highest in waste generation among Indonesian provinces; this paper investigates the demographic, economic, and environmental determinants of waste generation, specifically addressing the research question of how these factors vary across regencies. Novelty: This study extends previous waste generation studies by applying Geographically Weighted Regression (GWR) to the East Java context, initially considering demographic, economic, and environmental variables, and identifying spatial variations in the significant determinants of waste generation. Research Methods: Secondary data from 35 regencies/cities in 2023 were analyzed using GWR with a Bisquare Fixed kernel, which was selected as the optimal weighting function compared to Fixed kernels and OLS. Finding/Results: Surabaya City recorded the highest waste generation, while the GWR model achieved a goodness-of-fit of 92.72%, higher than the multiple linear regression model. The results confirm that the influence of waste generation determinants is not uniform across regions, indicating significant spatial heterogeneity in East Java.

Dimension Reduction of Socioeconomic Factors in Deforestation Analysis in Indonesia Using Sparse PCA

Mitha Rabiyatul Nufus — 2026-06-30

Introduction/Main Objectives: Deforestation remains a major environmental challenge in Indonesia under diverse socio-economic conditions. This study applies Sparse Principal Component Analysis (SPCA) to identify the key socio-economic variables associated with deforestation patterns. Background Problems: Analyses of deforestation drivers often involve numerous correlated variables, leading to multicollinearity and making interpretation difficult. Therefore, an approach is needed to reduce data dimensionality while retaining the most relevant information. Novelty: This study employs SPCA to simultaneously perform dimensionality reduction and variable selection, producing a more interpretable framework for identifying socio-economic factors related to deforestation at the provincial level in Indonesia. Research Methods: Provincial-level socio-economic data from Statistics Indonesia were analyzed using SPCA to address multicollinearity and derive interpretable components. Spatial autocorrelation was assessed using Moran’s I. Finding/Results: SPCA reduced the variables into two interpretable components and identified six key contributing variables while excluding three with limited influence. Moran’s I values for the first (0.402) and second (0.258) sparse principal components indicated significant positive spatial clustering of provinces with similar deforestation-related characteristics. Research Limitations: The analysis is limited to provincial-level secondary data and may not fully capture local-scale variations or all determinants of deforestation.

Examining the Local Effects of Food Security Index Components Across Kalimantan Using Geographically Weighted Regression

Meirinda Fauziyah — 2026-06-30

Introduction/Main Objectives: Food security remains a critical concern across Kalimantan Island, where substantial spatial disparities exist among its 56 regencies and cities, making conventional global regression models inadequate for capturing localized differences. Background Problems: This study addresses the limitation of Multiple Linear Regression in accounting for spatial heterogeneity in the relationships between Food Security Index components and the overall index, raising the question of which components exhibit spatially varying local effects across locations. Novelty: This study presents the first spatially explicit analysis of food security determinants at the regency and city level across Kalimantan, employing Haversine distance combined with adaptive Gaussian kernel weighting within GWR a combination not previously applied in this context. Research Methods: GWR was applied to cross-sectional 2024 data from the Food Security and Vulnerability Atlas, incorporating Cross Validation bandwidth selection and Weighted Least Squares parameter estimation. Finding/Results: The GWR model outperformed MLR with an R² of 59.63% and MSE of 38.5241. The ratio of population per health worker and average years of schooling for women were the most spatially dominant components, significant in 45 and 43 locations respectively, supporting the need for location-specific policy interventions across Kalimantan.

Mapping and Modeling Crime Factors in North Sumatra Using GWGPR

Eva Kosasih — 2026-06-30

Introduction/Main Objectives: Crime remains a significant social issue influenced by socio-economic factors and exhibiting spatial variation, particularly in North Sumatra Province, which recorded the highest number of criminal cases in Indonesia in 2024. This study aims to identify significant factors affecting crime and examine the spatial variation of their effects across districts/cities in North Sumatra. Background Problems: Global regression models often fail to capture crime patterns due to overdispersion and spatial heterogeneity, leading to inconsistent relationships across regions. Novelty: This study employs Geographically Weighted Generalized Poisson Regression (GWGPR), which simultaneously addresses overdispersion and spatial heterogeneity, providing a more robust localized analysis than global models. Research Methods: Using secondary data from 33 districts/cities in North Sumatra, the variables include population density, open unemployment rate, mean years of schooling, and Gini ratio. The analysis involves Poisson regression,dispersion testing,Generalized Poisson Regression, spatial heterogeneity testing, and GWGPR. Finding/Results: The significant factors affecting crime are the open unemployment rate, mean years of schooling, and population density, while the Gini ratio is not significant. Limitation: This study is limited by the use of data covering only the year 2024 and a limited set of socio-economic variables, which may not fully capture all factors associated with crime.

Classification of Village Development Status in Bekasi Regency Using Ensemble Learning and SMOTE-Based Class Balancing

Ridwan Mochamad Ridwan — 2026-06-30

Introduction/Main Objectives: This study aims to classify village development status in Bekasi Regency using machine learning based on the 2024 Village Potential Statistics (PODES) and the Village Development Index (IDM). Background Problems: Conventional descriptive assessments ignore complex socio-economic relationships, and class imbalance further reduces model predictive performance. Novelty: This study integrates PODES data, ensemble learning, and SMOTE to improve classification, providing a reliable, data-driven framework for village profiling and planning. Research Methods: Following preprocessing and a 70:30 split, SMOTE was applied to the training data, and four tree-based models (Decision Tree, Bagging, Random Forest, XGBoost) were evaluated using standard classification metrics. Finding/Results: The Random Forest model combined with SMOTE achieved the best classification performance, with an accuracy of 0.7778 and consistently high AUC values across all classes. The most influential predictors were the dominant economic sector, number of farmer groups, availability of basic health services, and presence of micro-business units. These findings demonstrate that combining ensemble learning with SMOTE improves village development classification and provides valuable support for evidence-based rural development planning in Bekasi Regency.

Analyzing Medium and Long Text Indonesian Tourism Feedback Using Topic Modeling and Sentiment Analysis

Sulisetyo Puji Widodo — 2026-06-30

Introduction/Main Objectives: Tourism is a vital sector supporting Indonesia’s economic growth, making the effective utilization of public feedback essential for improving service quality. Most feedback is collected through web-based forms in the form of open-text responses that provide rich insights but remain underutilized due to their unstructured nature. Background Problems: This study examines the challenge of identifying the most suitable topic modeling and sentiment analysis techniques for analyzing medium- and long-text feedback in the Indonesian tourism context. Novelty: The novelty lies in the comparative evaluation of classical topic modeling algorithms against modern embedding-based approaches combined with multiple Indonesian transformer models, which has not been extensively explored in tourism-related datasets. Research Methods: The research compares LDA and NMF with BERTopic, Top2Vec, kBERT, and kUSE using coherence scores, and evaluates sentiment analysis using majority voting across transformer architectures. Finding/Results: The results show that BERTopic performed best for medium-length text, while NMF was optimal for long text, and a RoBERTa-based model achieved the highest sentiment agreement. Positive sentiment often appeared in feedback on facilities and fees, whereas negative sentiment dominated topics on environmental and governance issues. These findings offer valuable insights for tourism managers and policymakers in prioritizing improvements and refining strategies.

Ensemble Boosting Models for Forecasting Rice Prices in Indonesia

Muhammad Jimmy Saputra — 2026-06-30

Introduction/Main Objectives: Rice is a key staple commodity influencing food security and inflation in Indonesia, making accurate price forecasting essential. In this study, we aim to compare ensemble boosting models and identify the best-performing model for rice price prediction. Background Problems: Notably, rice prices exhibit non-linear patterns over time, while classical statistical methods have limitations in capturing such complexities, resulting in suboptimal forecasting performance. Novelty: This study proposes a lag-based approach that uses lag variables as the only predictors, arranged across multiple input schemes to flexibly capture historical patterns without external variables. Research Methods: Daily national medium rice price data (Jan 2021–Jan 2026) from the National Food Agency are modeled using Gradient Boosting Machine (GBM) and LightGBM, with hyperparameter tuning via Optuna. The forecasting framework relies exclusively on significant lag variables without incorporating exogenous factors. Model performance is evaluated using RMSE, MAE, and MAPE. Findings/Results: LightGBM with optimized hyperparameters achieves the best performance (RMSE = 66.389; MAE = 50.213; MAPE = 0.362%). Furthermore, forecasts for the next 89 days indicate stable prices around Rp13,360–Rp13,395/kg, with no significant fluctuations.