Speaker: Richard Lockhart (SFU)
Location: Earth Sciences Building ESB 5104, 3pm – 4pm
Cancelled
Unfortunately, we have had to postpone this talk until a later date. We apologize for any inconvenience this may cause.
Abstract
Penalized regression methods permit data analysts to fit models with more adjustable parameters than data points by imposing strong prior assumptions on the relationships between variables. As a result classical tools of uncertainty assessment no longer work. I have been involved in one approach to uncertainty assessment for problems of this sort. I will try to summarize the issues and contrast the current approaches.
