Forecasting Oil Price by Hierarchical Shrinkage in Dynamic Parameter Models

<div>Forecasting performance measurement of models with LASSO prior on constant and time-varying coefficients, <i>h</i> = 1. (a) is the absolute forecast errors, (b) is the forecast errors squared, and (c) is the log predictive likelihood.</div>

Discrete Dynamics in Nature and Society

fig1

Figure 1

Figure 1: Forecasting Oil Price by Hierarchical Shrinkage in Dynamic Parameter Models 