Stock Price Prediction using Long-Short Term Memory and Temporal Convolutional Network

Predicting stock prices with high accuracy is difficult due to the fluctuating, complex, and chaotic nature of the stock market. The aim of this research is to address the challenges associated with stock price characteristics by proposing a solution through the integration of Long Short-Term Memory and Temporal Convolutional Network. The predicted data will be standardized using the MinMaxScaler function, ensuring that the utilized data is adjusted to fall within the 0 to 1 range. Furthermore, the separation of training data and testing data will be carried out with variations of 90, 10, and 80.10. Modeling will be carried out on 4 hybrid models namely TCN-LSTM, LSTM-TCN, TCAN-LSTM, and LSTM-TCAN. The test was carried out by comparing the 4 models using the same parameters and repeating the parameter training process 3 times. The test results on the 90.10 data variant showed that the superior TCN-LSTM model had the smallest average RMSE value of 0.03098 while the 80.20 data variant shows that the LSTM-TCN model is superior with an average RMSE value of 0.025605