PREDIKSI HARGA PENUTUPAN SAHAM BANK CENTRAL ASIA: IMPLEMENTASI ALGORITMA LONG SHORT-TERM MEMORY DAN PERBANDINGANNYA DENGAN SUPPORT VECTOR MACHINE

Abstract

Stock is an instrument of the financial market that is very popular among other instruments because it has an attractive yield. The research discusses the prediction of Bank Central Asia shares, named BBCA, using the Long Short-Term Memory (LSTM) method. The LSTM model is a very popular deep learning algorithm that is suitable for predicting time-related data, historical data, and sequential data. We configure the LSTM model with the following hyperparameters: number of neurons of 60, batch_size of 64, timesteps of 32, epoch of 12, and dense layer of one unit while the configuration for SVM support vector machine model with Gaussian Radial Basic Function kernel and hyperparameter γ = 0.0001 and c = 1000. BBCA prediction results are quite good when compared to the SVM model with a MAPE of 1.07%.

Keywords

• bbca, deep learning, mape, rnn, stocks
• saham