<![CDATA[Predicting stock prices is a challenging yet crucial task in financial markets. This research aims to compare the performance of two Long Short-Term Memory (LSTM) neural network models for forecasting the closing price of Nokia Corporation (NOK) stock: a univariate model using only historical closing prices and a multivariate model incorporating open, high, low, close, and volume (OHLCV) data. Utilizing historical daily data from 2015 to 2025, both models were trained to predict the next day's price based on the previous 60 days. The models' accuracy was rigorously evaluated using three key metrics: Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE). The findings revealed a decisive outcome. The univariate LSTM model consistently outperformed its multivariate counterpart across all evaluation metrics. It achieved an MAE of 0.0591, an RMSE of 0.0887, and a MAPE of 1.39%, while the multivariate model recorded higher values of 0.0623, 0.0934, and 1.45%, respectively. This study concludes that for NOK stock prediction, a simpler model with fewer features proved to be more effective. The additional data points in the multivariate model did not enhance predictive accuracy and may have introduced noise, suggesting that the historical pattern of closing prices alone is a more powerful predictor for this particular asset.]]>
Copyrights © 2025
