摘要：Detecting the number of breaks in the mean can be challenging when it comes to thelong memory framework. Tree-based procedures can be applied to time series when thelocation and number of mean shifts are unknown and estimate the breaks consistentlythough with possible overfitting. For pruning the redundant breaks information criteriacan be used. An alteration of the BIC, the LWZ, is presented to overcome long-rangedependence issues. A Monte Carlo Study shows the superior performance of the LWZ toalternative pruning criteria like the BIC or LIC
关键词：long memory; mean shift; regression tree; ART; LWZ; LIC
