STOCK PRICE PREDICTION USING ARTIFICIAL NEURAL NETWORKS WITH THE ADALINE METHOD

Abstract

Prediction of stock prices has an important role in making investment decisions in financial markets. In this study, researchers propose the use of artificial neural networks with the Adaline method to predict stock prices. The Adaline method is a learning algorithm that iteratively adapts network weights to minimize prediction errors. In the early stages, researchers collect relevant stock price historical data and process it into a training dataset. Next, build an artificial neural network with one input layer, one output layer, and use a linear activation function. The network weights are initialized randomly, and the training process begins. In network training, the researcher applies the Adaline algorithm with the descending gradient method to optimize the objective function. The researcher updates the network weights based on the difference between the actual stock price and the stock price predicted by the network. This process is repeated iteratively until it converges or reaches the specified stopping criteria. After the training is complete, the researcher evaluates the network performance using validation data that was not used in the training. Researchers used evaluation metrics, such as the mean absolute error (MAE) and root mean square error (RMSE), to measure the accuracy of network predictions. In addition, researchers also compared the performance of the Adaline network with other prediction methods, such as linear regression. The experimental results show that the artificial neural network with the Adaline method is able to provide accurate stock price predictions. Researchers observed a significant reduction in prediction error compared to the linear regression method. In addition, the Adaline network also shows good adaptability to changes in market trends. This research makes an important contribution to the development of stock price prediction methods using artificial neural networks. The results can be used as a guide for investors and stock traders to make more informed and effective investment decisions. Keywords : stock price prediction, artificial neural network, Adaline method, learning algorithm, performance evaluation.