Binary and Continuous Data, a Comparative Analysis using Machine Learning in Stocks

Abstract

The study included two distinct methods for addressing the ruling class, as well as ten mechanical hints culled from ten years of archival data as recommendation values. The second solution included converting signals to a binary dossier before employing the governing class, whereas the first arrangement involved altering the indications using stock market principles as a constant dossier. Each indicator model was evaluated using three verification systems that were constructed as input systems.

The assessment outcomes revealed that RNN and LSTM models outperformed other forecasting models by a wide margin over the full dossier. The open ocean education models outperformed the extra models for the double dossier; however, the difference was less evident since the models' performance improved significantly in the second design. Overall, this study demonstrated that machine intelligence and deep learning algorithms have the capacity to predict stock market currents, assisting financiers in making informed decisions and underestimating risks.