Claim Missing Document
Check
Articles

Found 3 Documents
Search
Journal : VARIANSI: Journal of Statistics and Its Application on Teaching and Research

PEMODELAN MULTIVARIATE ADAPTIVE REGRESSION SPLINE (MARS) PADA INDEKS HARGA SAHAM GABUNGAN (IHSG) TAHUN 2018 – 2023 Zakiyah Mar'ah; Ruliana, Ruliana; Magfirah Septiana
VARIANSI: Journal of Statistics and Its application on Teaching and Research Vol. 6 No. 01 (2024)
Publisher : Program Studi Statistika Fakultas MIPA UNM

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35580/variansiunm144

Abstract

Nonparametric regression is one of the methods used to estimate the pattern of the relationship between response variables and predictor variables where the shape of the regression curve is unknown and is generally assumed to be contained in an infinite dimensional function space and is a smooth function (Eubank, 1999). The MARS method is one method that uses a nonparametric regression approach and high-dimensional data. These namely data has a number of predictor variables of 3 ≤ k ≤ 20 and data samples of size 50 ≤ n ≤ 1000. This research discusses Multivariate Adaptive Regression Spline (MARS) Modeling on the Composite Stock Price Index (JCI) 2018 - 2023. MARS modeling is obtained from a combination of basis function (BF), maximum interaction (MI), and minimum observation (MO) based on the minimum Generalized Cross Validation (GCV) value. The results of this study were obtained from the combination value of BF = 16, MI = 1, and MO = 2 with GCV = 60710.98. The factors that affect the Jakarta Composite Index (JCI) are Inflation (X1), Rupiah to USD Exchange Rate (X3), and Money Supply (X4).
Perbandingan Metode ARIMA dan Single Exponential Smoothing dalam Peramalan Nilai Ekspor Kakao Indonesia Fahmuddin S, Muhammad; Ruliana, Ruliana; Mustika M, Sitti Sri
VARIANSI: Journal of Statistics and Its application on Teaching and Research Vol. 5 No. 03 (2023)
Publisher : Program Studi Statistika Fakultas MIPA UNM

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35580/variansiunm193

Abstract

Indonesia is a country with an open economy, one of the sources of foreign exchange needed by a country with an open economy is exports. Cocoa is one of Indonesia's main export commodities that makes an important contribution to the country's economy, but the value of Indonesian cocoa exports fluctuates, that is there are inconsistent changes from time to time. The purpose of this study is to determine the results of forecasting the value of Indonesian cocoa exports, as well as to determine the best method for forecasting. This research compares the ARIMA and Single Exponential Smoothing methods to determine the best forecasting method. The best method is selected based on the smallest MAPE value. Based on the results of data analysis, the best forecasting model using the ARIMA method is the ARIMA (1, 0, 1) model, which has a MAPE value of 10.38060%. Meanwhile, the best forecasting model using the Single Exponential Smoothing method is with α = 0.16, which has a MAPE value of 10.92874%. So that the best method for forecasting the value of Indonesian cocoa exports is the ARIMA method
Metode Radial Basis Function Neural Network Untuk Klasifikasi Kab/Kota Tertinggal Di Provinsi Sulawesi Selatan Ruliana, Ruliana; Rais, Zulkifli; Mar'ah, Zakiyah; Hasnita, Hasnita
VARIANSI: Journal of Statistics and Its application on Teaching and Research Vol. 5 No. 03 (2023)
Publisher : Program Studi Statistika Fakultas MIPA UNM

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35580/variansiunm197

Abstract

A disadvantaged area is an area that has the characteristics of tending to be left behind compared to other areas. Radial basis function neural networks are a part of Artificial Neural Networks, which use radial basis activation functions and are commonly used in classification cases. All districts/cities in South Sulawesi province have different characteristics from other districts/cities. Therefore, districts/cities are grouped into 2 groups to identify districts/cities that have characteristics that tend to be the same based on indicators of regional underdevelopment. The grouping results are then used as actual values ​​for classification using the RBFNN method, to determine the classification results and performance of the RBFNN method. In classifying districts/cities in South Sulawesi province based on indicators of regional underdevelopment using the radial basis function neural network method, an accuracy value of 91% was obtained using a comparison of 55% training data and 45% testing data and an f-measure value of 92% was obtained