<![CDATA[ Stock price is the stock price that occurs in the stock market at a certain time determined by market participants. Stock prices fluctuate up and down from time to time, most investors only use instinct to predict stock prices. Therefore, it is necessary to analyze the time series to predict stock prices. One of the models in time series is ARIMA. However, the ARIMA model has the disadvantage of only being able to follow a linear time sequence pattern. to address these weaknesses, ANN models are used that can follow nonlinear patterns. This is the ARIMA-ANN hybrid. The purpose of this study is to identify the model and determine the accuracy of the Arima-ANN hybrid model and predict the stock price in the next period. The Data used is secondary data from the site investing.com. determination of ARIMA model is done by dividing the data into training data and testing data. The results obtained by the best ARIMA model is ARIMA (2,1,2) with MAPE 1.468% of the training data. With the best ARIMA model, obtained residuals for ANN input. The result obtained is the best network architecture 5-10-1 learning rate 0.3 with the smallest error value. From the test results of the hybrid model ARIMA-ANN to the testing data obtained MAPE value of 7.024%. Then the prediction for the next 25 days obtained an average of 7915 rupiah per day and MAPE 6.349%.Keywords: Hybrid ARIMA-ANN, Predictions, Stock]]>
Copyrights © 2022
