Date and Time
March 2, 2005
1:30 PM - 3:00 PM

Availability
Open to the public
No RSVP required

Many critical questions in medicine require the analysis of complex multivariate data, often from large datasets describing numerous variables for numerous subjects. The multivariate nature of these data can make it difficult to assess the associations of the predictors and the outcomes of interest. The purpose of this talk is to present CoPlot, a novel method of graphical analysis of multivariate data.

CoPlot is based on multidimensional scaling (MDS). However, unlike MDS, CoPlot simultaneously presents the associations among both the observations and the variables. In doing so, CoPlot enables the user to identify clusters of observations, to identify variables that are correlated, and to identify observations that are particularly dominant in a particular variable or set of variables.

For this Research in Progress seminar, the presenters will discuss several examples of the use of CoPlot for exploratory analyses of multivariate data including its use in identifying outlier observations, selecting variables for subsequent regression analysis, and as a tool for assessing publication bias in meta-analysis, among others.