StockTM: Accurate Stock Price Prediction Model Using LSTM

Mohammad Diqi
DOI: https://doi.org/10.35842/ijicom.v4i1.50
2022-10-16
International Journal of Informatics and Computation
Abstract:Stock prediction aims to forecast a future stock price trend to assist investors in making strategic investment choices. However, it is hard to predict the price in dynamic conditions, which causes investors hard to anticipate equities because of the unstable prices. Thus, in this paper, we present a novel stock price prediction model based on the Long Short-Term Memory (LSTM) algorithm. Several steps are taken in creating a stock prediction model, including collecting datasets, pre-processing, extracting features, training and validating the model using evaluation metrics techniques. Based on the experimental results, the proposed prediction model can obtain good accuracy with a small error rate in an extensive dataset training. Therefore, it can be a promising solution to deal with the dynamic prices. Moreover, the proposed model can achieve the results obtained: RMSE EMA10 of 0.00714, RMSE EMA20 of 0.00355, MAPE EMA10 of 0.07705, and MAPE EMA20 0.05273.
What problem does this paper attempt to address?