E.D. Asriani, - and Sugiman, - and P. Hendikawati, - (2016) Estimasi Multivariate Adaptive Regression Splines (MARS) Pada Indeks Harga Saham Gabungan (IHSG). Unnes Journal of Mathematics, 5 (2). pp. 170-180. ISSN 2460-5859

PDF - Published Version Download (353kB)

Abstract

The purpose of this study is to know: (1) a estimation best MARS on CSPI with criteria GCV; (2) importance predictors variables against the model best obtained. Variabels affecting Composite Stock Price Index (CSPI) are inflation, interest rate, exchange rate the Rupiah againts the u.s.dollar, Dow Jones index, Nikkei 225 index, and Hang Seng index. MARS model is derived by combination of BF, MI, and MO through trial and error. MARS method on CSPI because nonparametric and high dimention is data has variabels predictors from 3 to 20 and data sampel from 50 to 1000. The analysis MARS method on CSPI with do testing parameters of regression nonparametric model, standaritation, and The results estimation MARS best on CSPI is BF=18, MI=1, and MO=1, GCV minimum is 0,05640. Predictors variables that were significans are inflation; exchange rate the rupiah againts the US$; Dow Jones index; interest rate; and Nikkei 225 index with contributions of importance are 100%; 86,54114%; 84,31259%; 38,18755%; and 32,75410%.

Actions (login required)

View Item