Online Program Home
  My Program

Abstract Details

Activity Number: 353 - Contributed Poster Presentations: Section on Nonparametric Statistics
Type: Contributed
Date/Time: Tuesday, August 1, 2017 : 10:30 AM to 12:20 PM
Sponsor: Section on Nonparametric Statistics
Abstract #322952
Title: Pruning for Nonparametric Change Point Analysis
Author(s): Wenyu Zhang* and David S Matteson and Nicholas James
Companies: Cornell University and Cornell University and Cornell University
Keywords: Dynamic Programming ; Incomplete U-Statistics ; Multivariate time series ; Pruning

Change point analysis is a statistical tool to attain homogeneity within time series data. We propose a pruning approach for nonparametric estimation of multiple change points. This general purpose change point detection procedure, cp3o, approximates the goodness-of-fit metric and applies a pruning step to the dynamic program to greatly reduce the search space and computational costs. A large class of existing goodness-of-fit change point objectives can immediately be utilized within the framework. We further propose two algorithms by incorporating two popular nonparametric goodness-of-fit measures with cp3o. e-cp3o uses E-statistics, and ks-cp3o uses Kolmogorov-Smirnov statistics. The only distributional assumption that e-cp3o makes is that the absolute alpha-th moment exists, for alpha in (0, 2). It can be used for both univariate and multivariate time series, to detect any type of distributional change. ks-cp3o makes no distributional assumptions, but is restricted to the univariate case.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2017 program

Copyright © American Statistical Association