Jeff Tian's Research in Data Analysis and Risk Management
Risk Identification and Management Techniques
I am interested in the applicability and effectiveness of
various statistical analysis techniques and models
in analyzing software engineering data,
particularly, various risk identification and management methods
that have many utilities in software management and improvement.
Specific techniques that I have investigated or plan to investigate
include:
- regression (linear, non-linear, logistic, etc.) models,
- principle component analysis and factor analysis,
- discriminant analysis and decision algorithms,
- tree-based modeling (both regression and classification trees),
- pattern matching approach to data analysis,
- abducted reasoning and other AI-based techniques,
- artificial neural network models,
- mathematical programming and optimization models.
Many such techniques, particularly tree-based models,
have been used by us to analyze diverse software engineering data,
including
product and process metrics,
customer surveys, and product defects and
reliability
data.
Such analyses have proved to be effective
in identifying problematic areas for focused improvement actions.
This investigation is summarized in the following paper:
-
J. Tian,
"Risk Identification Techniques for Defect Reduction and Quality Improvement",
Software Quality Professional,
Vol.2, No.2, pp.32-41, March, 2000.
Software Prototyping
Software prototypes can be viewed as an inexpensive probe to evaluate
the risk and uncertainties associated with software development to
collect information to allow us to make informed decisions.
Some of my work in this area is summarized in the following paper:
-
S. Cardenas, J. Tian and M.V. Zelkowitz.
"An Application of Decision Theory for the Evaluation of Software Prototypes".
Journal of Systems and Software,
Vol.19, No.1, pp.27-39, Sept., 1992.
Prepared by Jeff Tian
(tian@engr.smu.edu).
Last update: May 2, 2003.
Back to Jeff Tian's home page