Article Per Year (5 Year)
p-Index From 2021 - 2026
0.778
P-Index
This Author published in this journals
All Journal EduMatika: Jurnal MIPA Economic: Journal Economic and Business
Ardicha Appu Sianturi
Universitas Negeri Medan
Chemistry Decision Sciences, Operations Research & Management Economics, Econometrics & Finance Education Mathematics Physics

Published : 4 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 4 Documents
Search

 1 
Integrating garch, markowitz mean–variance, and lstm for risk, volatility, and portfolio analysis of four major Indonesian banking stocks 2015 – 2024 Risca Octaviyani Hutapea; Gizka Triyunita Sinaga; Ardicha Appu Sianturi
Economic: Journal Economic and Business Vol. 5 No. 1 (2026): ECONOMIC: Journal Economic and Business
Publisher : Lembaga Riset Mutiara Akbar (LARISMA)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.56495/ejeb.v5i1.1401

Abstract

This study examines the risk structure, volatility behavior, and optimal portfolio construction of four major banking stocks in Indonesia BBCA, BBRI, BMRI, and BBNI during the 2015–2024 period through the integration of the GARCH(1,1) model, Markowitz Mean–Variance optimization, and Long Short-Term Memory (LSTM)-based forecasting. Daily closing price data is transformed into log returns and tested for stationarity before further analysis. The GARCH estimation results indicate persistent high volatility across all stocks (?+? close to 1), with BBNI and BBRI the most responsive to market shocks, while BBCA remains the most stable stock. Markowitz optimization produces a Minimum Variance portfolio dominated by BBCA, while the Maximum Sharpe portfolio allocates funds entirely to BBCA due to its superior return efficiency. The LSTM is able to represent price trends well, as evidenced by low prediction error values for BBRI and BBNI and accuracy between 58–61 percent. The integration provides a comprehensive analytical framework for understanding changing market risk dynamics and supporting adaptive investment decision-making in the Indonesian banking sector. These findings confirm that the hybrid approach can improve risk mapping while maximizing portfolio performance through a combination of historical information, dynamic volatility, and price trend predictions.
Integration of survival analysis in predicting customer churn risk to optimize life insurance redemption value formulation Fachriz Effendy K; Risca Octaviyani Hutapea; Ardicha Appu Sianturi
Economic: Journal Economic and Business Vol. 5 No. 2 (2026): ECONOMIC: Journal Economic and Business
Publisher : Lembaga Riset Mutiara Akbar (LARISMA)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.56495/ejeb.v5i2.1580

Abstract

The risk of voluntary policy surrender is a major threat to the liquidity and stability of life insurance companies' premium reserves. Conventional actuarial valuations, which assume a static depreciation rate, often fail to accurately mitigate this risk. This study aims to integrate a predictive analytics approach into actuarial mathematics to create a dynamic Surrender Value formula. Using 10,000 historical observations of financial customers as a proxy, the analysis was conducted using the Kaplan-Meier estimator and Cox Proportional Hazard (Cox PH) regression. The Kaplan-Meier estimation results show that the probability of policy survival experiences an exponential decay from 95.8% in the first year to 80.2% in the fifth year. Cox PH modeling confirms that entry age (hazard ratio = 1.048) and female gender (hazard ratio = 1.535) significantly increase the surrender risk, while active customer interaction (hazard ratio = 0.589) acts as a protective factor. The resulting cumulative individual hazard probabilities are then integrated as weighting constants into the surrender charge formula. This integration produces penalty recommendations that adapt to each policyholder's risk profile, providing more proportional and equitable liquidity protection for insurance company operations.
Comparison of Cox Proportional Hazards and Weibull Regression Models in Survival Analysis of Heart Failure Patients Using UCI Repository Data Ardicha Appu Sianturi; Risca Octaviyani Hutapea; May Rani Tabitha Sinaga
EduMatika: Jurnal MIPA Vol. 5 No. 3 (2025): EduMatika: Jurnal MIPA
Publisher : Lembaga Riset Mutiara Akbar

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.56495/emju.v5i3.1327

Abstract

Heart failure is a leading cause of death worldwide, with a high mortality rate due to decreased heart function and systemic complications. Survival analysis is used to understand factors that influence patient survival and estimate the risk of death based on clinical characteristics. This study aims to analyze factors that influence survival time in heart failure patients and compare the performance of the Cox Proportional Hazards (CoxPH) model with the Weibull Accelerated Failure Time (AFT) in predicting the risk of death. Data are from the Heart Failure Clinical Records Dataset (UCI Repository) which includes 299 patients with variables such as age, anemia, hypertension, serum creatinine levels, and ejection fraction. The analysis was performed using the Kaplan–Meier, CoxPH, and Weibull AFT methods with evaluation through AIC and C-index values. The results show that age, anemia, hypertension, and creatinine increase the risk of death, while ejection fraction is protective. The CoxPH model performed better (AIC 958.46; C-index 0.741) than the Weibull AFT (AIC 1282.24; C-index 0.259). Therefore, CoxPH is recommended for estimating relative risk between patients, while Weibull AFT is more suitable for estimating absolute survival duration.
Modeling inflation in indonesia using the adf stationarity test, garch, and random forest regression based on monthly consumer price index data from 2014 to 2024 Ardicha Appu Sianturi; Risca Octaviyani Hutapea; Gizka Triyunita Sinaga; Enjelita Simangunsong
EduMatika: Jurnal MIPA Vol. 5 No. 3 (2025): EduMatika: Jurnal MIPA
Publisher : Lembaga Riset Mutiara Akbar

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.56495/emju.v5i3.1426

Abstract

Inflation is one of the key macroeconomic indicators that reflects price stability and overall economic conditions of a country. This study aims to examine the characteristics, volatility, and forecasting of monthly inflation in Indonesia over the period 2014–2024 by employing descriptive statistical analysis, time series approaches, and machine learning methods. The descriptive results indicate that Indonesia’s inflation has an average rate of 1.71% with a standard deviation of 1.32, a minimum value of 0.04%, and a maximum value of 8.36%. The inflation distribution is asymmetric, as evidenced by a positive skewness of 1.88 and a kurtosis value of 5.15, suggesting the presence of extreme inflation spikes during certain observation periods. Time series analysis reveals that the inflation data do not fully satisfy the stationarity assumption at the level form, thus requiring differencing prior to further modeling. Volatility estimation using the GARCH model indicates the existence of inflation volatility persistence. Furthermore, forecasting results using the Random Forest method demonstrate that the model is able to closely track actual inflation patterns under normal conditions, although predictive accuracy declines during periods of heightened volatility. Overall, this integrated approach provides a more comprehensive understanding of the dynamics of inflation in Indonesia.
Co-Authors Enjelita Simangunsong Fachriz Effendy K Gizka Triyunita Sinaga May Rani Tabitha Sinaga Risca Octaviyani Hutapea Risca Octaviyani Hutapea