Application of Extreme Learning Machine Method on Stock Closing Price Forecasting Pt Aneka Tambang (Persero) TBK

Abstract

Artificial neural networks are modeling methods that can capture complex input and output relationships. This method is widely used in forecasting and classification. However, in its application, there are some disadvantages in terms of low learning rate resulting in computational delay. Extreme Learning Machine (ELM) was introduced to overcome these problems. This method is believed to be able to produce more accurate forecasting results with a low level of forecasting error. In Indonesia, stocks are one of the most popular investments for investors. Stock prices tend to be volatile which is influenced by the amount of market supply and demand, so forecasting analysis is needed to minimize the risks that may occur. This research applies the ELM method to forecast the closing price of PT ANTM Tbk shares from January 1, 2018 - October 31, 2022. The data used is secondary data obtained from the official website https://id.investing.com. The ELM method is applied by dividing training data for ELM modeling and testing data used in the forecasting process. The model architecture of the ELM method uses a combination of inputs obtained from the PACF plot, hidden nodes with a range of 5-50, and one output layer. The results of this study show excellent forecasting accuracy in terms of forecasting. This is measured by the MAPE value of 0.0358. The architecture formed in the ELM process is one input, 31 hidden nodes, and one output. Forecasting the closing price of PT ANTM Tbk shares with 1-31-1 architecture produces a forecasting value that shows a low decline, but is quite stable.