A survey on visualization in predictive modeling

PhD Qualifying Examination


Title: "A survey on visualization in predictive modeling"

by

Mr. Yuanzhe Chen


Abstract:

Since many applications such as the analysis of customer behavior, disease 
and drug effectiveness are predictive by nature, predictive modeling lies 
in a very important position in various research areas including 
marketing, bioinformatics, finance and so on. However, designing and 
building an accurate and effective predictive model is far from trivial. 
Many phases of the model building process, such as sample selection, 
feature selection, parameter tuning, model comparison and validation, are 
tedious and difficult to optimize. Moreover, sometimes for a single model, 
the optimized parameter settings may differ from those used in various 
applications and thus require the assistance of domain experts who are not 
familiar with computer algorithms. Information visualization techniques 
are powerful for observing and exploring massive amounts of data, and 
could hence provide many valuable visual guidance for users during the 
model building process. In this paper, we first briefly introduce the 
general framework of predictive modeling, and then give a comprehensive 
review of the visualization techniques used in predictive modeling. At the 
end of this survey, we point out some potential research directions in the 
future.


Date:			Wednesday, 7 October 2015

Time:                  	9:30am - 11:30am

Venue:                  Room 4475
                         Lifts 25/26

Committee Members:	Prof. Huamin Qu (Supervisor)
 			Prof. Dit-Yan Yeung (Chairperson)
 			Prof. Chi-Keung Tang
 			Dr. Raymond Wong


**** ALL are Welcome ****