Dara wrote:

> John this is what I could do to help out for that talk:

> Make you fancy interactive CDF to explain the concepts for the talk, assumption being you have a good machine to run it on.

Thanks! But first we need to do the research for this talk. If you can help David Tweed do [the project he proposed](http://forum.azimuthproject.org/discussion/1445/possible-gpgpu-task-linearbilinear-regression-on-el-nino-dataset/?Focus=12127#Comment_12127), that would be great. I believe I can fill in details about what datasets to use (the surface air temperatures in a certain rectangle of points) and what quantities we want to predict (the El Niño 3.4 index). I'm hoping David Tweed or you will ask me for these details.

I would prefer to let him, and Daniel Mahler, and you, be the experts on the _methods_ of prediction. I would like to start by some using "standard" or "routine" methods of prediction, just to see what happens. I hope you guys can explain those methods to me, or point me to explanations.

Later, perhaps after December, we can try more creative and interesting things.