<![CDATA[PT. Bank Central Asia is a private company that has superior shares in the Lq45 category but has share prices that fluctuate every period. So forecasting is needed to predict stock prices in the next period. These fluctuations can cause linear and nonlinear relationships in historical stock price data. This research uses the Hybrid ARIMA-NN approach, where the ARIMA model is able to overcome data non-stationarity while the Neural Network is used to capture nonlinear patterns that cannot be explained by the ARIMA model by using the residuals as NN input, the hybrid model can increase forecasting accuracy. The data used is weekly data on closing stock prices for the period January 2019 to June 2024. Prediction measurements use Mean Absolute Percentage Error. The research results show that forecasting with Hybrid ARIMA(2,1,2)-NN(1-5-1) obtained a MAPE value of 3.99% smaller than the ARIMA(2,1,2) a MAPE value of 4.13%, that the accuracy of the forecasting model is good.]]>
Copyrights © 2024
