Just a summary of some cool things I have done in Leeds so far. I helped somebody with a parameter problem by creating a map of the results of a machine learning clustering algorithm in parameter space- that was fun. I found that negative Chi values worked well for the OPTICS algorithm (though I am not sure if this is meaningful or useful.

The training datasets were clouds. Optics was used to cluster, for example:

Clusters seemed to default to -1 (No cluster) and were not productive. There were only two parameters (min cluster size and Xi), so I made a parameter space and set up a comprehensive search, with the following results for the number of clusters and the size of largest cluster:

You can see that the algorithm may be productive if Xi values are negative.

In addition, on the cryosphere I generated a map of glacier velocities (and edited some wikipedia articles; it would appear that there are two gargoyle ridges in Antarctica! Very strange.)

Everyone else did it as well, but I am pleased by it regardless. Using SNAP.

That’s about it for training so far- looking forward to getting back.

Sam

Training: Leeds / Space: weather or not by blogadmin is licensed under a Creative Commons Attribution CC BY 3.0