Lorenzo Pascual

Universidad Carlos III of Madrid

Visiting the University of South Carolina


Bootstrap Predictive Inference for Autoregressive Processes:
A Simpler Approach


In this talk, a new resampling scheme for the construction of bootstrap prediction intervals in AR(p) processes is presented. I will discuss the relevance of fixing the last p observations as well as using the backward representation in order to obtain replicates of the available series. A simulation study shows the good behavior of the new approach compared to existing methods. Ideas for future research such as extensions to more complex models, are given.


Back to Colloquium Series