cover
Article Per Year (5 Year)
All
Home Page
OAI Link
Editorial Team
Contact
Reviewer
Google Scholar
Contact Name
Dr. Muhammad Ahsan
Contact Email
muh.ahsan@its.ac.id
Phone
+6281331551312
Journal Mail Official
inferensi.statistika@its.ac.id
Editorial Address
Department of Statistics Faculty of Science and Data Analytics Institut Teknologi Sepuluh Nopember (ITS) Kampus ITS Keputih Sukolilo Surabaya Indonesia 60111
Location
Kota surabaya,
Jawa timur
INDONESIA
Inferensi
Published by Institut Teknologi Sepuluh Nopember Surabaya
ISSN : 0216308X     EISSN : 27213862     DOI : http://dx.doi.org/10.12962/j27213862
Core Subject : Science, Education, Social, Engineering,
Computer Science & IT Decision Sciences, Operations Research & Management Engineering Mathematics Social Sciences
The aim of Inferensi is to publish original articles concerning statistical theories and novel applications in diverse research fields related to statistics and data science. The objective of papers should be to contribute to the understanding of the statistical methodology and/or to develop and improve statistical methods; any mathematical theory should be directed towards these aims; and any approach in data science. The kinds of contribution considered include descriptions of new methods of collecting or analysing data, with the underlying theory, an indication of the scope of application and preferably a real example. Also considered are comparisons, critical evaluations and new applications of existing methods, contributions to probability theory which have a clear practical bearing (including the formulation and analysis of stochastic models), statistical computation or simulation where the original methodology is involved and original contributions to the foundations of statistical science. It also sometimes publishes review and expository articles on specific topics, which are expected to bring valuable information for researchers interested in the fields selected. The journal contributes to broadening the coverage of statistics and data analysis in publishing articles based on innovative ideas. The journal is also unique in combining traditional statistical science and relatively new data science. All articles are refereed by experts.
Arjuna Subject : Matematika - Statistik dan Probabilitas
Articles 157 Documents
 12   13   14   15   16 
Comparison of Ordinal Logistic Regression and Artificial Neural Network in Stunting Prevalence Classification Risnawati, May; Fathurahman, M.; Prangga, Surya
Inferensi Vol 8, No 3 (2025)
Publisher : Department of Statistics ITS

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j27213862.v8i3.22287

Abstract

The prevalence of under-five stunting in one of the crucial health problems in Indonesia. Stunting is a growth and development disorder in children due to chronic malnutrition and repeat infections that can have a negative impact on children’s physical and cognitive development. This study aims to analyse the accuracy of the classification of the prevalence of stunting on regencies/cities in Indonesia, in 2022 using two methods, namely Ordinal Logistic Regression (OLR) and Artificial Neural Network (ANN). OLR is development of logistic regression applied to response variables with more the two categories that have levels or ranks, while ANN is a method that mimics the function of the biological nervous system and is designed for complex information processing. This study used two proportions of data splitting namely 80:20 and 90:10. Each method produce two models, OLR 1 and OLR 2 for the OLR method, and ANN 1 and ANN 2 for the ANN method. The results show that the ANN 1 model with 80:20 data proportion performs better than other models with an accuracy of 63.37%.
Comparison of ARIMA, LSTM, and Ensemble Averaging Models for Short-Term and Long- Term Forecasting of Non-Stationary Time Series Data Pratiwi, Windy Ayu; Sumertajaya, I Made; Notodiputro, Khairil Anwar
Inferensi Vol 8, No 3 (2025)
Publisher : Department of Statistics ITS

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j27213862.v8i3.22643

Abstract

This study aims to forecast the highest weekly selling rate of the Indonesian Rupiah (IDR) against the US Dollar (USD) and identify the most accurate model among ARIMA, LSTM, and Ensemble Averaging. The evaluation results indicate that ARIMA achieves an accuracy of 97.75%, demonstrating strong performance in short-term forecasting, while LSTM achieves 99.98% accuracy, excelling in capturing complex and dynamic patterns in long-term predictions. The Ensemble Averaging approach attains the highest accuracy of 99.99%, proving to be the optimal solution by combining ARIMA’s stability with LSTM’s adaptability, resulting in more precise and stable predictions. The findings of this study highlight that the ensemble approach is more effective than individual models, as it balances accuracy and prediction stability across various forecasting scenarios. This method serves as a reliable tool for addressing market volatility and contributes significantly to the advancement of financial and economic forecasting techniques that are more adaptive and accurate.
Small Area Estimation of Child Poverty on Java Island In 2021 (Comparison of EBLUP and Hierarchical Bayes) Istiana, Nofita; Tanur, Erwin; Ubaidillah, Azka; Sitanggang, Yuliana Ria Uli; Nainggolan, Rosalinda
Inferensi Vol 8, No 3 (2025)
Publisher : Department of Statistics ITS

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j27213862.v8i3.23311

Abstract

Information about child poverty is very important to ensure that children get their rights. Indonesia's decentralized system requires child poverty data in each district/city. Data provision at this level is constrained by a non-specific sample design used for certain age groups, so the sample age group for children is not always sufficient for each district/city. Therefore, direct estimation produces a high relative standard error (RSE), so it requires small area estimation (SAE). SAE that is often used is EBLUP, which assumes that the variable of interest is normally distributed. Child poverty data does not meet the normality assumption, so SAE with Hierarchical Bayes with Beta distribution (HB Beta) is proposed in this study. The result is direct estimation, EBLUP, and HB Beta produce relatively similar estimated values, but HB Beta has the lowest RSE.
Earthquake Point Clustering Using Self Organizing Maps (SOM) In Sulawesi and Maluku Regions Irwan, Irwan; Zaki, Ahmad; Widiyaningrum, Eka Janivia
Inferensi Vol 8, No 3 (2025)
Publisher : Department of Statistics ITS

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j27213862.v8i3.23331

Abstract

Earthquakes pose a major threat in Indonesia, particularly in complex tectonic regions like Sulawesi and Maluku. To support disaster mitigation, this research employs the Self Organizing Maps (SOM) method—an unsupervised technique that reduces data dimensionality into an intuitive two-dimensional form—to cluster earthquake data using four key variables: longitude, latitude, magnitude, and depth. The dataset includes 5,275 earthquake records from 2022, sourced from the Meteorology, Climatology, and Geophysics Agency (BMKG). SOM training produced 25 neurons, which were then grouped into three optimal clusters using hierarchical clustering, validated by internal metrics: the lowest Connectivity Index (296.1512), highest Silhouette Index (0.3304), and a Dunn Index of 0.0058. Cluster 1, with 13 neurons, covers eastern Sulawesi and Maluku, featuring medium magnitude and depth. Cluster 2, with 11 neurons, represents central to southern Sulawesi, characterized by low magnitude and shallow depth. Cluster 3, comprising a single neuron, includes western regions with high-magnitude, very deep earthquakes. Keywords⎯ Clustering, Earthquake, Internal Validation, Self Organizing Maps (SOM).
Forecasting Indonesia's Non-Oil and Gas Exports Using Facebook Prophet: A Seasonal and Trend Analysis Erfiani, Erfiani; Wijaya, Ferdian Bangkit
Inferensi Vol 8, No 3 (2025)
Publisher : Department of Statistics ITS

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j27213862.v8i3.23337

Abstract

This study aims to analyze and predict the trend of Indonesia's non-oil and gas exports using the Facebook Prophet model, focusing on identifying seasonal patterns, trends, and volatility present in the export data. Monthly export data from 2015 to 2025, sourced from the Statistics Indonesia (BPS), were used as the basis for analysis. The dataset revealed notable seasonal patterns and substantial volatility, particularly in the period following 2020. To model these dynamics, three Prophet model configurations were tested: one considering only annual seasonality, combining both annual and monthly seasonality, and another incorporating only monthly seasonality. The evaluation of these models showed with an initial Mean Absolute Percentage Error (MAPE) of 8.70%. This model was then optimized through hyperparameter tuning. The optimal parameter configuration (changepoint_prior_scale = 0.5, seasonality_prior_scale = 0.01, fourier_order = 3) resulted in a significant improvement, reducing the MAPE to 4.73%. This optimized model demonstrated its capacity to more precisely capture the complex patterns. Furthermore, the study projected Indonesia’s non-oil and gas exports for the period from April 2025 to December 2026. The projections indicate a relatively stable export trend within the range of 20,000 to 22,000 million USD per month, with consistent seasonal patterns.
Spatial Extreme Value Analysis of Extreme Rainfall Using the Extremal-t Process Nuroini, Husna Mir'atin; Sutikno, Sutikno; Purhadi, Purhadi
Inferensi Vol 8, No 3 (2025)
Publisher : Department of Statistics ITS

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j27213862.v8i3.23351

Abstract

Indonesia’s diverse topography, consisting of coasts, lowlands, highlands, and mountains, results in a wide range of weather and climate conditions, enabling various hydrological phenomena such as extreme rainfall, hurricanes, high temperatures, and storms. In recent years, global warming has emerged as a major environmental concern, with one of its significant impacts being climate change. This, in turn, increases the frequency and intensity of extreme hydrological events, potentially causing floods, transportation and communication disruptions, infrastructure damage, agricultural losses, and threats to human life. This study aims to identify the best model and estimate the return levels of extreme rainfall in Ngawi Regency from March 1990 to November 2022 using spatial extreme value analysis with max-stable processes and the extremal-t process. Daily rainfall data from 1990 - 2018 were used for model training, while data from 2018 - 2022 were allocated for model testing to validate predictive performance. Parameter estimation was conducted using Maximum Likelihood Estimation (MLE) and Maximum Pairwise Likelihood Estimation (MPLE), solved through the Broyden-Fletcher-Goldfarb-Shanno (BFGS) Quasi-Newton numerical iteration method. The analysis shows that the best trend surface model has average rainfall and variance influenced by latitude, while the distribution shape is unaffected by latitude or longitude, indicating isotropy. Furthermore, the return level prediction demonstrates higher accuracy when applied over a three-year period.
CART and Random Forest Analysis on Graduation Status of Halu Oleo University Students Rahman, Gusti Arviana; Notodiputro, Khairil Anwar; Sartono, Bagus; Surimi, La
Inferensi Vol 8, No 3 (2025)
Publisher : Department of Statistics ITS

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j27213862.v8i3.23336

Abstract

Classification and Regression Tree (CART) is a popular classification method and it is used in various fields. The method is capable to be applied on various data conditions. An alternative method of CART is random forest. These two methods of classification were studied in this paper using graduation data of Halu Oleo University. This data was interesting due to the imbalance problem existed in the data. We compared several scenarios, namely the CART and Random Forest methods, Random Forest with oversampling, and Random Forest with undersampling. There were three explanatory variables considered in the model including Study Program, GPA, and TOEFL score. The results showed that the best method to classify the student’s graduation status at Halu Oleo University is Random Forest without handling imbalanced data, as it provided the highest sensitivity. This suggests that Random Forest, even without specific adjustments for data imbalance, can effectively capture the patterns in the data and provide accurate classifications, making it a robust choice for this dataset.
Geographically Weighted Negative Binomial Regression for Modeling Maternal Deaths in East Java Province Calista Deva Salfatah; Purhadi
Inferensi Vol 9 No 1 (2026)
Publisher : Department of Statistics ITS

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j27213862.v9i1.7811

Abstract

The number of maternal deaths in Indonesia increased from 3,572 in 2022 to 4,460 in 2023. East Java, as one of the provinces with the largest population in Indonesia, still faces challenges in reducing the number of maternal deaths, with 499 cases recorded, placing it in second place with a contribution of 11.19% of the total national cases. This study aims to model the number of maternal deaths in East Java using the Geographically Weighted Negative Binomial Regression (GWNBR) method, which considers spatial aspects and overdispersion of data. The GWNBR model, both with and without exposure, was applied using the Adaptive Bisquare, Gaussian, and Tricube Kernel weighting functions. The exposure variable represented by the number of pregnant women to adjust for inter-regional risk. Based on the model goodness-of-fit (AICc), the GWNBR model with exposure using the Adaptive Gaussian Kernel weighting function was the best model. This model identified four groups of districts/cities based on the similarity of significant predictor variables. Significant variables across all regions are the ratio of community health centers and hospitals per 100,000 population, while the percentage of obstetric complication treatment and postpartum family planning participants are only significant in some regions.
Zero Inflated Bivariate Ordered Probit on Poverty Levels in Eastern Indonesia Andriano; Purhadi; Ismaini Zain
Inferensi Vol 9 No 1 (2026)
Publisher : Department of Statistics ITS

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j27213862.v9i1.8668

Abstract

This study examines the development of a bivariate ordinal probit regression model with zero-inflated effects. The bivariate ordinal probit regression method that can overcome zero-inflated effects is the Zero Inflated Bivariate Ordered Probit (ZIBOPR) method. ZIBOPR is a statistical method for examining the relationship between predictor variables and two correlated response variables that have levels and zero-inflated effects. A theoretical study was conducted to obtain parameter estimates for the ZIBOPR model using the Maximum Likelihood Estimator (MLE) method. The estimation equation obtained from MLE produces a non-closed form equation, so it is continued with the Bern, Hall, Hall, and Hausman (BHHH) numerical iteration method. The resulting parameters are then tested simultaneously with the Likelihood Ratio Test (LRT) and individually using the Wald test. The ZIBOPR model was applied to the case of the percentage of poor people and the poverty depth index in 176 districts/cities in Eastern Indonesia in 2023 using five predictor variables, namely life expectancy, per capita expenditure, average length of schooling, Labor Force Participation Rate (LFPR), and Infant Mortality Rate (IMR). The results showed that the ZIBOPR model parameters could be estimated using MLE and followed by BHHH numerical iteration. The resulting parameters could then be tested simultaneously using LRT and individually using the Wald test. Subsequently, ZIBOPR modeling with poverty rate levels and poverty depth index levels showed that the five parameters in the predictor variables had a significant effect on poverty rate levels and poverty depth index levels. The results of the best model selection analysis using the vuong test show that the ZIBOPR model better models the percentage of poor population and the poverty depth index than the bivariate ordinal probit regression model.
Penalized Multivariate Adaptive Regression Splines with Generalized Cross Validation for Modeling Health Insurance Ownership in East Java Deby Victoria; Addina Nurkamila; Yanuar Ibnu Ridho; Rafly Tawekal; Dita Amelia
Inferensi Vol 9 No 1 (2026)
Publisher : Department of Statistics ITS

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j27213862.v9i1.8747

Abstract

The average programmatic health insurance coverage is a key indicator of public welfare and the effectiveness of healthcare policies. This study proposes the use of Penalized Multivariate Adaptive Regression Splines (PMARS) with Generalized Cross Validation (GCV) to model the determinants of this average coverage rate across 38 regencies and cities in East Java Province, using secondary data from the 2025 BPS East Java publication. Several PMARS specifications are evaluated by varying polynomial degrees, numbers of basis functions, and penalty values to identify the optimal model structure. The PMARS approach effectively captures nonlinear relationships, interaction effects, and variable importance within a flexible regression framework. The optimal model is a quadratic specification consisting of 16 basis functions, a maximum interaction of two, and an optimal penalty parameter, explaining 94.46% of the variability in the average health insurance coverage with a minimum GCV value of 5.49654. Access to improved sanitation, clean water, and health complaints are identified as the most influential determinants, all exhibiting positive associations that drive the need for formal financial protection. These results demonstrate the effectiveness of the GCV-PMARS methodology for modeling complex socio-health data and provide empirical insights to support policies aimed at strengthening universal health coverage, ensuring equitable programmatic utilization, and advancing Sustainable Development Goal 3 on Good Health and Well-Being.

Page 15 of 16 | Total Record : 157

 12   13   14   15   16 

Filter by Year

2018 2026


Reset
Filter By Issues
All Issue Vol 9 No 1 (2026) Vol 8, No 3 (2025) Vol 8, No 2 (2025) Vol 8, No 1 (2025) Vol 7, No 3 (2024) Vol 7, No 2 (2024) Vol 7, No 1 (2024) Vol 6, No 2 (2023) Vol 6, No 1 (2023) Vol 5, No 2 (2022) Vol 5, No 1 (2022): Inferensi Special Issue: Seminar Nasional Statistika XI 2022 Vol 4, No 2 (2021): Inferensi Vol 4, No 1 (2021): Inferensi Vol 3, No 2 (2020): Inferensi Vol 3, No 1 (2020): Inferensi Vol 2, No 2 (2019): Inferensi Vol 2, No 1 (2019): Inferensi Vol 1, No 2 (2018): Inferensi Vol 1, No 1 (2018): Inferensi More Issue