Kyoung-jae Kim talks with
ScienceWatch.com and answers a few questions about
this month's Fast Moving Front in the field of
Engineering.
Article: Financial time series forecasting using
support vector machines
Authors: Kim,
KJ
Journal: NEUROCOMPUTING, 55 (1-2): 307-319 SEP 2003
Addresses: Dongguk Univ, Coll Business Adm, Dept Informat
Syst, 3-26 Pil Dong, Seoul 100715, South Korea.
Dongguk Univ, Coll Business Adm, Dept Informat Syst, Seoul
100715, South Korea.
Why do you think your paper is highly
cited?
This paper represented one of the earliest studies on financial forecasting
using support vector machines (SVMs), a powerful algorithm based on
statistical learning theory. SVMs are usually applied to engineering
problems, such as pattern recognition, but this paper applied SVMs to a
financial problem.
"...the main theme of my Ph.D.
thesis is financial forecasting using various
data mining techniques..."
In addition, this paper compared SVMs with other popular data mining
techniques, including case-based reasoning and backpropagation neural
networks. The paper also performed the McNemar test; a statistical
procedure used to compare two proportions which are dependent or
correlated, and is one of the most popular nonparametric tests for
statistical significance, to validate the generalization of the
experimental results. These may be among the primary reasons for the high
citation rate.
Does it describe a new discovery, methodology, or
synthesis of knowledge?
This study is the first attempt to apply SVMs to real-world financial
forecasting.
Would you summarize the significance of your paper
in layman's terms?
SVMs are the most popular data mining techniques. This study applied SVMs
to predict real-world stock price indices. In addition, this study examined
the feasibility of applying SVMs in financial forecasting by comparing it
with backpropagation neural networks and case-based reasoning.
How did you become involved in this research and
were any particular problems encountered along the way?
My research interests include the application of data mining techniques to
business problems, such as financial forecasting, corporate bond rating,
credit rating, and customer classification. Thus, the main theme of my
Ph.D. thesis is financial forecasting utilizing various data mining
techniques.
I have previously published several research papers on financial
forecasting using data mining techniques, whose subjects include artificial
neural networks, case-based reasoning, support vector machines, and hybrid
models of multiple techniques.
Kyoung-jae Kim, Ph.D.
Associate Professor
Department of Management Information Systems
College of Business Administration
Dongguk University
Seoul, Korea
KEYWORDS: ARTIFICIAL NEURAL-NETWORKS; INDEX
FUTURES.