| dc.description |
<p>This package contains the input datasets, experimental results, and source codes developed during the scientific initiation research project entitled <i>Study of Economic Variables for Stock Market Trend Prediction via Machine Learning</i>, funded by FAPESP and conducted at the Universidade Estadual de Campinas (UNICAMP).</p>
<p>The following files are included:</p>
<ul>
<li>File <b>Data_ExperimentalInputs.zip</b>: contains all the input datasets used throughout the research. This includes historical stock price data (obtained via the Yahoo Finance API), macroeconomic indicators (e.g., SELIC, IPCA, PIB), and microeconomic indicators (e.g., ROE, P/L, VPA) used as features in the machine learning models. This data is already preprocessed and ready for use.</li>
<li>File <b>Data_ExperimentalResults.zip</b>: includes the complete experimental results, such as model evaluation metrics (MAE, MSE, RMSE, R²), prediction graphs, comparison tables of different strategies (including the proposed method and the buy-and-hold approach), and return simulations.</li>
<li>File <b>SourceCodes.zip</b>: contains all the Python source codes used in the study, structured into the following subfolders:
<ul>
<li><i>CodigosExperimentoFeatures</i>: scripts and notebooks implementing the main experiments using different combinations of input features (price only, macro, micro, and combined).</li>
<li><i>ExperimentoKfold</i>: same experiments as above, but with K-Fold cross-validation to assess the generalization capacity of the models.</li>
<li><i>ExperimentosPadrao</i>: baseline models trained only with previous stock prices, used as a reference point for performance comparisons.</li>
<li><i>CodigosTrade</i>: scripts used to simulate the investment strategy based on the model forecasts and to calculate portfolio performance.</li>
<li><i>Resultados</i>: output files generated by the code, including result tables, CSVs, and graphical analyses.</li>
</ul></li>
</ul>
<p>This organization ensures full transparency and reproducibility of all the steps of the research, from data preparation to each strategy evaluation.</p> |
|