PREDIKSI HARGA SAHAM HARIAN MENGGUNAKAN JARINGAN SYARAF TIRUAN (JST) DENGAN ALGORITMA PROPAGASI BALIK

Abstract

Prediction of stock prices is useful for investors to understand how investments is suitable into organization in the future. Predictions can anticipate fluctuations in stock prices and can also help investors to make decisions. Artificial Neural Networks (ANN) provides a fast and flexible way to predict stock prices, and show better results compared to conventional methods. The algorithm used for prediction of the stock is backpropagation with data model of time series. This algorithm is a supervised training method that serves to minimize errors on output generated by the network. The results of this study indicate that the optimum results obtained from the use of Artificial Neural Networks is a combination of the activation function Logsig-Purelin, with the level of value Mean Absolute Percentage Error (MAPE) is 2.43% and the correlation coefficient is 94.08%, so that the algorithm is feasible and effective for stock price prediction