International Journal of Computer Applications |
Foundation of Computer Science (FCS), NY, USA |
Volume 185 - Number 21 |
Year of Publication: 2023 |
Authors: Priyank Gupta, Sanjay Kumar Gupta, Rakesh Singh Jadon |
10.5120/ijca2023922935 |
Priyank Gupta, Sanjay Kumar Gupta, Rakesh Singh Jadon . Stock Market Prediction using RNN-based Models with Random and Tuned Hyperparameters. International Journal of Computer Applications. 185, 21 ( Jul 2023), 12-17. DOI=10.5120/ijca2023922935
The stock market in India has become more passionate in recent years. Because of the maneuverability of the stock market, it is also difficult to predict future trends and patterns in the stock market. Various Deep Learning (DL) methods, such as Recurrent Neural Networks (RNN), produce excellent results in stock market forecasting. In this paper, we integrate RNN-based models such as Long-Short-Term Memory (LSTM), Stacked LSTM, Gated Recurrent Unit (GRU), Stacked GRU, Bidirectional LSTM, Bidirectional GRU, and a Hybrid model to predict the Moving Average Convergence Divergence (MACD) of National Stock Exchange (NSE) of India, i.e., NIFTY50 market index. Some hyperparameters are also considered when training these models, as these hyperparameters control the behavior and performance of such models. Two experiments are carried out to train these RNN-based models: manually selecting and tuning hyperparameters. Metrics such as Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Squared Percent Error (MAPE) are used to assess performance. Both experimental results show that the Bidirectional GRU model is the most effective at predicting MACD values in India's NIFTY50 stock market index.