PEMODELAN REGRESI SPLINE TRUNCATED UNTUK DATA LONGITUDINAL (Studi Kasus : Harga Saham Bulanan pada Kelompok Saham Perusahaan Penyedia Jasa Telekomunikasi Periode Januari 2009-Desember 2016)

Abstract

Stocks are securities that can be bought and sold by an individual or business entity as a token of ownership of a person or business entity in a company or limited liability company. Judging from market capitalization, stocks are divided into 3 groups: big capital, mid-cap, and small-cap. Stock prices fluctuated up and down due to the influence of several factors, one of which inflation. One of the stock prices that experienced ups and downs namely Telecommunications Service Provider Company. Telecommunication Service Provider Company is a company that provides telecommunication services to meet the needs of telecommunications. Longitudinal data are observations made by n mutually independent subjects with each subject observed repeatedly in different time periods that are mutually dependent. The longitudinal data modeling of stock price is done by spline truncated nonparametric regression approach for each observation subject. The best spline model depends on the determination of the optimal knot point, which has the minimum GCV value. The best spline truncated regression model lies in the 2nd order with 3 knot points for each subject in the longitudinal data. Prediction using the model in data out sample yield MAPE value for each subject is 8,75% for PT X., 17,47% for PT Y., and for 22,17% PT Z.