Dynamické priemerovanie modelov a predikčné modely makroekonomickej dynamiky
| Title in English | Dynamic model averaging and predictive models of macroeconomic dynamics |
|---|---|
| Authors | |
| Year of publication | 2024 |
| Type | Special-purpose publication |
| MU Faculty or unit | |
| Citation | |
| Description | The main objective of the thesis is to analyse inflation forecasts through dynamic or simple averaging over different countries and horizons. The study also investigated the effect of widening and sliding windows on the predictions. Among the methods of the predictions themselves, regularization techniques were found to be the most accurate. XGboost and Random Forest seemed to be the least accurate. Averaging the predictions in some cases was able to refine the inflation predictions against the individual prediction methods. The key innovation was Dynamic Model Averaging, which was able to minimize prediction errors. |
| Related projects: |