Pemodelan Metode Arima dan Support Vector Regression (SVR) Pada Harga Saham PT Telkom Indonesia

Abstract

Forecasting is an activity or process of predicting future values based on past data. In this study the data to be used is the stock price data of PT Telkom Indonesia. Forecasting that is commonly used in time series data is the ARIMA method, but there is a more practical method because it does not require the assumption of stationary data, namely the Support Vector Regression (SVR) method. The results of accurate stock forecasting will be used by investors to hedge potential market risks and provide opportunities for market speculators and arbitrage to profit from trading indices. So we need a system to predict the price movements of these stocks to help investors carry out appropriate analysis and action. The result is that risks can be minimized and profits can be increased. Although the SVR method has the advantage of not requiring stationary data assumptions, the SVR method also has a weakness, namely it is not good at time series data. The results of forecasting PT Telkom stock price data in September 2022 using SVR has a model with a MAPE of 3% this indicates that forecasting using the ARIMA method is better than the SVR method.