deux
- Obtenir le lien
- X
- Courriel
- Autres applications
- Obtenir le lien
- X
- Courriel
- Autres applications
sss
Here’s a shorter and clearer version of the table in your image: --- Key Improvements in ASW 5Y Prediction Model 1. Feature Selection Optimization Used automated tools to identify the most impactful variables. Focused the model on features aligned with the current market regime. 2. Refined Training Data (Post-2023 Focus) Excluded 2021-2022 data with abnormally high ASW levels. Trained the model on post-2023 data to better reflect today’s macroeconomic landscape and reduce bias. 3. Improved Data Frequency Handling Shifted from weekly to daily data for more precise modeling. Adjusted frequency conversion to minimize lag effects in ASW predictions. 4. Incorporation of New Features Added new variables from BPM research, improving sensitivity to sharp market movements and volatility. 5. Expanded Model Testing Tested 9 additional models beyond Random Forest: Tree-Based: XGBoost, LightGBM Ensemble: Stacking, Blending Neural Networks: GRU (RNN) Statistical Models: SARIMA --- This version keeps...
Commentaires
Publier un commentaire