UROP Project

Change-Point Detection in Time Series

Contact

Name

Daniel Holder

Program Director UROP

Telephone

workPhone
+49 241 80-90695

E-Mail

Key Info

Basic Information

Project Offer-Number:
389
Category:
UROP International, RWTH UROP
Field:
Mathematics
Faculty:
1
Organisation unit:
Chair of Stochastics and Institute of Statistics
Language Skills:
English or German
Computer Skills:
Programming skills, preferably in C and R.

MoveOn

Present day data sets are usually dependent time series streams which arrive sequentially, e.g. when monitoring data from financial markets, environmental data or digitized signals in engineering. There is great interest to detect structural changes in such time series in order to trigger measurements such as risk adjustments of financial positions, an alarm when monitoring for biosurveillance or switching to another channel with a lower data transfer rate when transmitting data through a multiplex channel. It is of great interest to study and compare the power of detection procedures to detect various kinds of changes in a data stream.

Task

After reading paper(s) and related material, the task is to implement, test and analyze detection methods we are studying in our group. Implementations and simulations will be done in R and P, respectively. P is our HPC compiler for R with extensions for parallel programming.

Requirements

Background in probability theory/statistics and basic knowledge of time series models. Knowledge of nonparametric statistics would be