The recent thread on biocalculus was quite interesting. The course below presents plots as well as other methods to make sense of multivariate data. It focuses on practical, often distribution-free methods that work in a wide variety of situations.
Untangling Multivariate Relationships a 2-day course on applied multivariate statistics Nov 1-2, 2007 $895 Colorado School of Mines - more info at: www.practicalstats.com Environmental measurements often are simultaneously recorded as a series of variables or species. Multiple chemicals are analyzed. Multiple measures of ecosystem health are recorded. If statistics are computed on each, one by one, the interrelations among variables, species and locations cannot be seen. Valuable information is lost. Multivariate methods resolve what at first may look like noise into recognizable patterns, providing new insight into the field of study. However, these methods are daunting to many scientists, with acronyms like PCA, CCA or CANOCO, and with unfamiliar terms like varimax rotation and detrended correspondence. The learning curve seems steep, and useful procedures go unused. UMR covers the multivariate methods of primary interest to environmental science, focusing on what each method is designed to do, when to use them, and when not to. Methods for simplifying and plotting data (ordination) are contrasted with those for establishing connections between and among source and response variables. Capabilities of various software packages are reviewed. Example data sets are analyzed by each student in class. Our goal is that students leave with a clear understanding of which method to use, and how to use it, for their own data. -- ----------------------------------------------------------- Dennis R. Helsel, Ph.D. http://www.practicalstats.com Practical Stats 9278 S. Lark Sparrow Dr. Littleton, CO 80126-5232 [EMAIL PROTECTED]
