Hi all;

We're currently investigating the possibility of an online evaluation program.

I'd like feedback on the probability calculation(s) you would like.

I'm currently using a method devised by Dr. Alan Jones which eliminates the effect of clustering in the dataset by counting only predictions with matching quakes without regard to the number of quakes in the windows.

Another method compares the quakes in prediction windows to the amount of time taken up by the predictions.

Does anyone have any other favorite methods they would like to see implemented? Mathematically valid methods, of course.

Roger