Skip to main content
eScholarship
Open Access Publications from the University of California

Analysis of Time Series Methods with Iterated Boosting

  • Author(s): Wang, Sean
  • Advisor(s): de Leeuw, Jan
  • et al.
Abstract

Datasets that have time elements should be analyzed with special models that account for the temporal aspect of the data. While there have been many methods developed to analyze time series data, the recent rise of data mining within the last two decades has opened up a whole new field to explore. Could data mining procedures be used to obtain even more accurate predictions for time series data?

Our focus here will be to compare a variety of methods that have been used to analyze time series data, along with a new one called iterated boosting, which takes the fitted values from boosting and treats them as the data to boost in the next iteration. We will test all the methods on crime data from the Los Angeles Police Department, gathered from 2000 to 2006.

When iterated boosting is applied to the data, the results performed about as well as the other methods, and did in fact outperform several of the methods.

Main Content
Current View