New LGB+ model enhances macroeconomic forecasting with linear and tree components

Researchers have developed LGB+, a novel boosting procedure designed to improve macroeconomic forecasting by incorporating a wider range of basis functions beyond traditional trees. This method evaluates both linear and tree-based models at each step, advancing the superior performer to enhance predictions, particularly in scenarios with strong linear dynamics or mixed linear-nonlinear signals. The approach allows for a native decomposition of forecasts into linear and nonlinear components, offering clearer insights into variable importance and historical influence. AI

IMPACT Introduces a new statistical method that could improve forecasting accuracy in economic applications.

RANK_REASON The cluster contains an academic paper detailing a new statistical method for forecasting. [lever_c_demoted from research: ic=1 ai=0.4]

Read on arXiv stat.ML →

• LGB+
• Philippe Goulet Coulombe

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →