John Sall (SAS Institute Inc.)
Title: From Big Data to Big Statistics
Now that we have Big Data we have new challenges. When big data is wide data there are too many things to look at. When we screen for just the big effects, we need to worry about selection bias. When we graph p-values, all the good ones bunch together near zero. When you have tall data, everything is significant, but many effects are too small to care about. When you have holes and bumps all over your data, you need something automatic to adapt to them. All this needs to be resolved in a Big Statistics analytic work flow.