Claim Missing Document
Check
Articles

PENGGUNAAN SUPPORT VECTOR REGRESSION DALAM PEMODELAN INDEKS SAHAM SYARIAH INDONESIA DENGAN ALGORITME GRID SEARCH Galih Hedy Saputra; Aji Hamim Wigena; Bagus Sartono
Indonesian Journal of Statistics and Applications Vol 3 No 2 (2019)
Publisher : Statistics and Data Science Program Study, SSMI, IPB University, in collaboration with the Forum Pendidikan Tinggi Statistika Indonesia (FORSTAT) and the Ikatan Statistisi Indonesia (ISI)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29244/ijsa.v3i2.172

Abstract

Indonesia as the largest Muslim population country in the world is a very potential market for sharia stocks. Sharia stocks performance can be seen from the Indonesia Sharia Stock Index (ISSI). Stock index modeling is conducted to determine the factors that affect the stock index or to predict the value of the stock index. Modeling using regression analysis is based on assumptions that do not always match with the characteristics of stock data that fluctuate. Support Vector Regression (SVR) method is a non-parametric approach based on machine learning. The problem often encountered in the analysis using SVR is to determine the optimal parameters to produce the best model. The determination of the optimal parameters can be solved by using the grid search algorithm. The purpose of this research is to make ISSI model using SVR with grid search algorithm with independent variable BI Rate, money supply, and exchange rate (USD / IDR). The best SVR model was obtained using weekly data with a total of 343 periods as well as a linear kernel with parameters ε = 0.03 and C = 2. The evaluation of the best model SVR is RMSE of 2.289 and correlation value of 0.873.
PENENTUAN NILAI AMBANG BATAS SEBARAN PARETO TERAMPAT DENGAN MEASURE OF SURPRISE Yumna Karimah; Aji Hamim Wigena; Agus Mohamad Soleh
Indonesian Journal of Statistics and Applications Vol 3 No 2 (2019)
Publisher : Statistics and Data Science Program Study, SSMI, IPB University, in collaboration with the Forum Pendidikan Tinggi Statistika Indonesia (FORSTAT) and the Ikatan Statistisi Indonesia (ISI)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29244/ijsa.v3i2.284

Abstract

Extreme rainfall can result in natural disasters such as floods and landslides. These natural disasters will cause damage and losses to the surrounding environment. Prevention of damage from natural disasters can be done by extreme rainfall estimation. Estimates of extreme rainfall are based on Generalized Pareto Distribution (GPD) which requires threshold value information. The threshold value can be determined by two methods, namely Mean Residual Life Plot (MRLP) and Measure of Surprise (MOS). The purpose of this study is to determine and compare the threshold values ​​of MRLP and MOS. The data used are 10-day and monthly rainfall data. The results of this study indicate that the procedure of MOS is shorter and easier than that of MRLP. Based on the cross validation result, the log-likelihood value of MOS is larger than that of MRLP, then MOS is better than MRLP.
The Impact of Using A Linear Model for the Ordinal Response of Mixture Experiments Utami Dyah Syafitri; Erfiani Erfiani; Agus M Soleh; Aji Hamim Wigena
ZERO: Jurnal Sains, Matematika dan Terapan Vol 9, No 2 (2025): Zero: Jurnal Sains Matematika dan Terapan
Publisher : UIN Sumatera Utara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30829/zero.v9i2.25760

Abstract

In a sensory test, the response is a Likert scale, which belongs to the ordinal scale. The ordinal response can be analyzed using a linear model approach; however, this approach can be misleading. This research aims to compare three different methods for ordinal response: the average score, the second-order Scheffe model, and the ordinal logistic model. The case study focused on the response to the taste of cookies resulting from the mixture experiment. The mixture experiment is one type of experimental design which is commonly used for product formulation. The research involved three ingredients with different lower bonds. The D-optimal design which also the {3,2} simplex-lattice design was chosen for the experiment. The three methods were conducted, and they all yielded the same results for the optimum composition; however, the ordinal model provided more information about the data's characteristics. The optimal formulation of each ingredient was 10%, 20%, 70%.