Institut für Informatik,
Knowledge Discovery from Tree-Ring Time Series
Tree-rings reflect climatic and environmental conditions in which
the trees have grown.
The study of the tree-rings can reveal the events such
as floods, forest fires, temperature changes, earthquakes,
etc. Besides for the archeology, the analysis of tree-rings time
series is used for studying the conditions and needs involved in the
growth of forests.
In this project we study the tree-ring time series using the
techniques and methods from the field of knowledge discovery and data
mining. The problems which are addressed in particular are the
discovery of characteristic patterns in tree-ring time series, and the
problem of dating tree-ring time series. The discovery of the
characteristics patterns can identify the timepoints of the specific
events which occurred during the existence of the sample trees. Dating
of tree-ring time series is used for determining the age of the sample
pieces of wood.
The algorithm for matching the sequences of values with the
set of time series has been developed in the frame of the study of
tree-ring time series. The algorithm was implemented using
C5.0 data mining tools and Perl programming language. The empirical
study of the use of matching algorithm for the above stated
problems of pattern recognition and dating has been done.
The work on knowledge discovery from tree-ring time series is
performed in the frame of a joint project of
Information Systems Group (DBIS) at Institute for Computer Science,
and Institute for
Forest Growth (IWW), Freiburg University.
- Prof. Georg Lausen
- Dr. Iztok Savnik
- Aldar Dougarjapov
- Prof. Heinrich Spiecker
- Dr. Hans-Peter Kahle
- Sebastian Hein
- Algorithm for Matching Sets of Time Series.
I. Savnik, G. Lausen, H.P. Kahle, H. Spiecker, S. Hein.
Technical Report, DBIS, Institut für Informatik,
Universität Freiburg, Mar 2000.
- MSTS: A System for Mining Sets of Time Series.
G. Lausen, I. Savnik, A. Dougarjapov.
Technical Report, Institut für Informatik, Universität
Freiburg, May 2000.
Updated: Fri Mar 3 16:48:24 MET 2000