Claim Missing Document
Check
Articles

Found 21 Documents
Search

Comparative Modeling of Pineapple Production Using Gaussian GLM and Random Forest Regression Radot MH Siahaan; Indah Gumala Andirasdini; Fuji Lestari; Dwi Mahrani; Amalia Listiani
ZERO: Jurnal Sains, Matematika dan Terapan Vol 10, No 1 (2026): Zero: Jurnal Sains Matematika dan Terapan
Publisher : UIN Sumatera Utara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30829/zero.v10i1.28721

Abstract

This study aims to conduct a comparative modelling of pineapple production at PT Great Giant Pineapple (GGP) using Gaussian GLM as parametric statistical approach and Random Forest Regression method as machine learning based on monthly data from 2014 to 2022. Multicollinearity testing and distribution fitting were conducted to validate the Gaussian assumption. For the Random Forest Regression, hyperparameters were optimized by tuning the number of trees (ntree) and the number of predictors at each split (mtry) with model stability evaluated using Out-of-Bag (OOB) error. The Gaussian GLM achieved a MAPE of 8.41% (R² = 0.106) for the GP3 clone and 11.27% (R² = 0.149) for the F180 clone. Random Forest Regression produced a testing MAPE of 9.28% (R² = 0.144) for GP3 and 12.11% (R² = 0.105) for F180. While both models achieved low prediction error based on MAPE, they differed in identifying influential variables and showed limited explanatory power as indicated by low R² values. The Gaussian GLM identifies air pressure as significant for both clones and rainfall for F180 clone, while Random Forest consistently identifies rainfall as the most influential predictor. These findings confirm the complementary strengths of parametric and machine learning approaches in supporting climate-based production planning and risk mitigation.
Black-Scholes Method for Rainfall Index-Based Agricultural Insurance Premiums Lestari, Fuji; Julianty, Dila Tirta; Vikarti, Maulida Magdalena
Enthusiastic : International Journal of Applied Statistics and Data Science Volume 6 Issue 1, April 2026
Publisher : Universitas Islam Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.20885/enthusiastic.vol6.iss1.art7

Abstract

Agriculture plays an important role in national economic development. However, it has the highest risk of loss due to its dependence on climate conditions. One of the efforts to reduce the risk of crop failure is through an agricultural insurance program. This study aimed to analyze the value of the rainfall climate index used and the calculation of agricultural insurance premiums based on it. The method used to determine the rainfall climate index was the historical burn analysis method, while the method used to calculate agricultural insurance premiums was the Black-Scholes method. The study showed significant spatial variation in rainfall index-based agricultural insurance premiums across Sumatra. Premiums rose with higher percentiles, with North Sumatra the highest (IDR 3.28–3.55 million) and Aceh the lowest (IDR 100–137 thousand). The inclusion of all rainfall stations revealed a more detailed spatial pattern. Overall, premiums strongly reflect local climatic conditions and can aid risk assessment and insurance planning.
Pemodelan Hybrid Radial Basis Function Neural Network Berbasis Ensemble Learning dalam Prediksi Parameter Nitrat (NO₃⁻) pada Sungai di Wilayah Tropis Rifka Noor Azizah; Azahra Sukesih; Isti Dina Febriyanti; Nur Faizaturrohmah; Fuji Lestari
Jurnal Kesehatan dan Pengelolaan Lingkungan Vol. 7 No. 2 (2026)
Publisher : Universitas Ahmad Dahlan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12928/jkpl.v7i2.16292

Abstract

Nitrate (NO₃⁻) concentration is one of the key indicators for evaluating river water quality, particularly in tropical regions with relatively high temperatures, which accelerate the nitrification process in the nitrogen cycle. Increasing human activities, such as residential development, agricultural practices, livestock farming, and aquaculture near rivers, can increase the nitrogen nutrient load, posing a risk of degrading water quality. Traditional methods such as the Pollutant Index (PI) and Water Quality Index (WQI) only reflect current conditions and cannot accurately predict changes in water quality. Therefore, this study aims to analyze the effectiveness of the Radial Basis Function Neural Network (RBFNN) model in predicting nitrate concentrations based on water quality parameters in rivers in tropical regions. This study was conducted by analyzing 164 publications providing data on water physico-chemical parameters, such as temperature, pH, TDS, TSS, DO, BOD, COD, nitrite, ammonia, and electrical conductivity. The obtained data were then processed through data cleaning, logarithmic transformation, feature engineering, and model training using an RBFNN combined with Gradient Boosting and Random Forest methods via an ensemble approach. Model evaluation was performed using the coefficient of determination (R²). The results of the study indicate that this model exhibits excellent predictive capability, with an R² value of 0.9907 for residential riverbanks and 0.9953 for agro-aquatic riverbanks, outperforming conventional ANN models in similar studies in Indonesia, which generally yield R² values in the range of 0.90–0.97, and comparable to the best RBFNN models ever reported in the international literature. Sensitivity analysis indicates that the parameter most influential on nitrate concentration is nitrite (40.25%), followed by electrical conductivity (19.80%) and TDS (11.10%). These results indicate that the RBFNN is effective in modeling nonlinear relationships among water quality parameters and has the potential to be developed as a core component in an early warning system for nitrate pollution, a decision support system for water resource managers, and a more efficient and adaptive data-driven water quality monitoring instrument to support the sustainable management of tropical rivers. Keywords: radial basis function, nitrate, tropical rivers, water quality, neural network
PERBANDINGAN PREDIKSI INDEKS HARGA SAHAM GABUNGAN (IHSG) MENGGUNAKAN METODE REGRESI RANDOM FOREST DAN EXTREME GRADIENT BOOSTING Agnes Monika Simalango; Muklas Rivai; Fuji Lestari
Jurnal Gaussian Vol 15, No 2 (2026): Jurnal Gaussian
Publisher : Department of Statistics, Faculty of Science and Mathematics, Universitas Diponegoro

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.14710/j.gauss.15.2.356-367

Abstract

Indeks Harga Saham Gabungan (IHSG) serves as the benchmark of the Indonesian stock market and is characterized by nonlinear and volatile movements driven by macroeconomic conditions, trading activities, and market sentiment. This study aimed to compare the forecasting performance of Random Forest and Extreme Gradient Boosting (XGBoost) for the IHSG using historical data from January 2019 to July 2025 with an 80:20 chronological train-test split. Model performance was evaluated using RMSE and MAPE, followed by variable importance analysis and Value at Risk (VaR) estimation to asses potential investment losses. The results showed that Random Forest outperformed XGBoost, achieving an RMSE of 117.54 and a MAPE of 0.90%, whereas XGBoost produced an RMSE of 136.45 and a MAPE of 1.03%. Variable importance analysis identified the highest index value as the most influential predictor, followed by the lowest index value, the opening index value, and trading volume. Furthermore, the estimated expected return was 0.00014, with VaR values of IDR 470,000.00, IDR 610,000.00, and IDR 960,000.00 at the 90%, 95%, and 99% confidence levels, respectively. Based on the evaluation results, Random Forest achieved better forecasting accuracy than XGBoost for the IHSG.
Pelatihan Olah Data dan Visualisasi Data Statistik dalam Peningkatan Kompetensi Perangkat Desa Badran Sari, Lampung Selatan Ayu Sofia; Fuji Lestari; Muklas Rivai; indah Gumala andirasdini; Dila Tirta Julianty
TeknoKreatif: Jurnal Pengabdian kepada Masyarakat Vol 5 No 1 (2025): TEKNOKREATIF : Jurnal Pengabdian kepada Masyarakat Volume 5 No 1
Publisher : Lembaga Penelitian dan Pengabdian Kepada Masyarakat (LP2M), Institut Teknologi Sumatera, Lampung, Indonesia.

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35472/teknokreatif.v5i1.1873

Abstract

The use of data processing skills is something that is very important in various fields. The data processing process can use various applications, one of which is the number/data processing application which is commonly known as the Ms. Excel application. Ms. Excel is a software program that allows users to process and calculate numerical data so that calculations and reading data are no longer done manually. The problem with partners is the lack of competence of village officials regarding the use of technology in processing village data. Based on the problems faced by partners, the PkM team offers a solution, namely providing training in processing and analyzing statistical data using Ms. Excel which aims to help village officials to be able to process data and be able to visualize the data into images/graphs that are more attractive to the community so that able to improve the quality of data processing contained in village officials.
Statistical Pattern Recognition of Lithosphere Anomalous Activity Along the Indonesian Ring of Fire Mika Alvionita S; Ardika Satria; Triyana Muliawati; Fuji Lestari; Danni Gathot Harbowo
Journal of Science and Applicative Technology Vol. 9 No. 1 (2025): Journal of Science and Applicative Technology June Chapter
Publisher : Lembaga Penelitian dan Pengabdian Masyarakat (LPPM), Institut Teknologi Sumatera, Lampung Selatan, Lampung, Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35472/jsat.v9i1.1850

Abstract

The introduction of statistical pattern recognition becomes highly important for assessing disaster threats such as earthquakes. This approach is significantly more comprehensive and suitable for long-term event forecasting. Therefore, in the future, efforts can be promptly made to reduce the risk of disasters resulting from anomalies in lithospheric activity, especially frequent earthquakes in the Sumatra Island region, Indonesia. Statistical pattern analysis of lithospheric activity anomalies can be categorized through classification. Earthquake classification is performed based on magnitude scale and mathematical calculations of earthquake parameter unit conversion. The classification method employed in this research includes machine learning methods like k-nearest neighbor and support vector machine. The evaluation metrics used for machine learning models are model accuracy and confusion matrix tables.
Perhitungan Cadangan Premi Asuransi Jiwa Berjangka dengan Menggunakan Metode Zillmer dan Fackler Kristiani Sitorus; Tiara Yulita; Fuji Lestari
Indonesian Journal of Applied Mathematics Vol. 4 No. 2 (2024): Indonesian Journal of Applied Mathematics Vol. 4 No. 2 October Chapter
Publisher : Lembaga Penelitian dan Pengabdian Masyarakat (LPPM), Institut Teknologi Sumatera, Lampung Selatan, Lampung, Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35472/indojam.v4i2.1944

Abstract

Human life is never free from risk. Along with the development of the era, humans began to realize the importance of protecting themselves in the event of a risk, including the risk of death. To overcome this, many individuals transfer the risk by registering themselves or their families with life insurance. The life insurance that is focused on is term life insurance, which is a form of protection with a certain period of time that has been set. In order to run its operations properly, insurance companies need to prepare premium reserves with accurate calculations. These calculations can be done through two actuarial methods, namely prospective and retrospective methods. This study uses the Zillmer method (prospective) and the Fackler method (retrospective) for calculating premium reserves. The purpose of this study was to determine the analysis of the results of calculating the premium reserve value from the two methods. The premium reserve values ​​calculated using the Zillmer and Fackler methods were different. At an interest rate of 6.25%, the Fackler method produces a higher premium reserve compared to the Zillmer method. This difference is because the Zillmer method includes other costs such as acquisition costs, administration, and agent commissions in calculating premium reserves. In contrast, the Fackler method does not take these costs into account in its calculations.
Perhitungan Premi Asuransi Pertanian Pada Tanaman Padi Di Provinsi Jawa Barat Menggunakan Metode Copula Fuji Lestari
Indonesian Journal of Applied Mathematics Vol. 5 No. 2 (2025): Indonesian Journal of Applied Mathematics Vol. 5 No. 2 October Chapter
Publisher : Lembaga Penelitian dan Pengabdian Masyarakat (LPPM), Institut Teknologi Sumatera, Lampung Selatan, Lampung, Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35472/indojam.v5i2.2435

Abstract

Agricultural insurance is an important instrument to protect farmers from risks that can threaten the sustainability of farming, such as crop failure due to natural disasters, pest attacks, and plant diseases. Rice, as a strategic commodity for national food security, is a priority in insurance protection programs. This study aims to calculate rice insurance premiums in West Java Province using the copula method, based on secondary monthly data of rice production and dry grain prices (GKG) from January 2020 to December 2023, as well as daily rainfall data. The methodology includes data exploration, distribution fitting, copula parameter estimation, and premium calculation based on rainfall threshold values. The analysis results indicate that the Gaussian copula is the best model for capturing the dependence between variables, with the lowest AIC and BIC values. Premium calculations based on rainfall indices show variations according to threshold values, with lower premiums for smaller thresholds, confirming the effectiveness of the Gaussian copula in modeling risk for agricultural insurance premiums.
Penentuan Nilai Premi Asuransi Pertanian Berbasis Indeks Curah Hujan Dengan Metode Simulasi Stochastic Weather Generator Menggunakan Distribusi Weibull Giandito Iskandar; Dila Tirta Julianty; Fuji Lestari; Elifas Hani Christien; Nia Ahmadia Sumardi; Welson Tawan; Natalie Evangelista; Debra Carol Lin G.N.
Indonesian Journal of Applied Mathematics Vol. 6 No. 1 (2026): Indonesian Journal of Applied Mathematics Vol. 6 No. 1 April Chapter
Publisher : Lembaga Penelitian dan Pengabdian Masyarakat (LPPM), Institut Teknologi Sumatera, Lampung Selatan, Lampung, Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35472/indojam.v6i1.2716

Abstract

The agricultural sector is vulnerable to various risks, which can lead to losses, such as rice farming that is susceptible to unstable rainfall. Therefore, agricultural insurance is needed to protect farmers from losses due to reduced harvests caused by various risks. The maximum value covered by the insurance company, called the indemnity value, reflects the total losses experienced by the farmers based on the average rice production yield. The average rice production yield is calculated using the estimated selling price obtained from the value- at-risk method amounting to IDR7,401,647.28/ton. This study aims to determine the premium that farmers must pay using a rainfall index value simulated using the Weibull Distribution with the Stochastic Weather Generator method. The result of this study is that the premium ranges from IDR143,567.5 – IDR11,511,142 or around 0.349% to 27.939% of the indemnity value amounting to IDR41,375,333.65.
Actuarial Pricing of Health Insurance for Breast Cancer Patients Using Markov Chain Method Fuji Lestari; Zalikha Alma
KUBIK Vol 11 No 1 (2026): KUBIK: Jurnal Publikasi Ilmiah Matematika
Publisher : Department of Mathematics, Faculty of Science and Technology, UIN Sunan Gunung Djati Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15575/kubik.v11i1.55191

Abstract

Breast cancer is a critical illness frequently experienced by women and requires high treatment costs. This can become a significant financial burden, especially if treatment continues over a long period. Health insurance serves as a financial protection solution to mitigate this risk. This study aims to calculate the net single premium for health insurance for breast cancer patients using the Markov chain method. A multistate model was built with five states: stage I, II, III, IV, and death. Patient transition data was obtained from the medical records of the Abdul Moeloek Regional General Hospital for the years 2023–2024. The results show that the net single premium increases with the extension of the coverage period. The net single premium for breast cancer patients over a 5-year period is . For periods of 6, 7, 8, 9, and 10 years, the premiums are 4,68 times, 9,08 times, 14,06 times, 19,48 times, and 25,25 times larger, respectively, compared to the net single premium for the 5-year period.premium.