This is the repository of the paper: "Boosting and Predictability of Macroeconomic Variables: Evidence from Brazil" that was published open access at Springer Nature - Computational Economics.
If you're interested in our results, we will guide you on the folder structure. If you have any questions, feel free to reach me via email here.
We use the boosting.Rproj file to automatically set the main folder as the working directory.
- code: here we store (almost) all the .R files.
- data: this folder contains everything we used for the development of the database. Including csv files of each variable, metadata and the code set up to directly get the data from Ipea API's.
- plots: here we stored all the plots generated. Some of them were used for the paper.
- tests: here we store all the .mat files generated by Ipeadata_predictability.R.
WIP
This paper aims to elaborate a treated data set and apply the boosting methodology to monthly Brazilian macroeconomic variables to check its predictability. The forecasting performed here consists in using linear and nonlinear base-learners, as well as a third type of model that has both linear and nonlinear components in the estimation of the variables using the history itself with lag up to 12 periods. We want to investigate which models and for which forecast horizons we have the strongest performance. The results obtained here through different evaluation approaches point out that, on average, the performance of boosting models using P-Splines as base-learner are the ones that have the best results, especially the methodology with two components: two-stage boosting. In addition, we conducted an analysis on a subgroup of variables with data available until 2022 to verify the validity of our conclusions. We also compared the performance of boosted trees with other models and evaluated model parameters using both cross-validation and Akaike Information Criteria in order to check the robustness of the results.
- Guilherme Schultz Lindenmeyer
- Prof. Hudson Torrent - Advisor
The study employs various approaches to assess the performance of the boosting models. We utilize P-Splines as base-learners and investigate the two-stage boosting technique. Furthermore, a subgroup of variables with data available until 2022 is analyzed to validate our conclusions.
The findings reveal that, on average, boosting models using P-Splines as base-learners, especially the two-stage boosting method, demonstrate superior performance. Cross-validation and AIC are used to evaluate model parameters, ensuring the robustness of the results. Comparisons are drawn against other models to provide a comprehensive assessment of the boosting methodology's effectiveness.
We can say that the application of the models obtained solid results for all the models applied, but, on average, the two-stage boosting modeling had the highest performance. Thus, we can say that this model is the best candidate among the models used here to forecast Brazilian macroeconomic variables with their own history.
We followed and extended the methodology originally introduced by Boosting nonlinear predictability of macroeconomic time series. Our code is adapted and extended from Timo Virtanen's original code.