Hi Jason,

I still have the same issues as described in my emails from 30^th August and 14^th September (please see below). This is now becoming a bit pressing for me, so I would like to ask whether you have the time now to look into the problem.

Kind regards,

Markus

 

 

Hi Markus,

 

I’m very sorry to not have responded. I am completely occupied with completing a two year project that is due at the end of this month (30 September). I think this problem can be fixed but can’t immediately determine how to fix it, so I have to defer the investigation. Please contact me again October 1 or later and I will look at it then.

 

Best,

Jason

 

 

Hi Jason,

I just wanted to follow up regarding my query. As mentioned in a previous email, your suggestion of deleting ‘fpr’ in the second field didn’t work. Below are the geoprocessing results.

I hope you could have a look at this issue.

Best wishes,

Markus

 

 

 

Hi Markus,

 

Thanks for your interest in MGET. I can help you with this but am on holiday so my ability to do so is limited until next week. I looked at the code and can see that the low-level part that produces the plot was designed to allow just one performance measure but the part that produces the ArcGIS input form might not. I have to do some experiments to know for sure and am not able to at the moment.

 

What happens when you delete the ‘fpr’ that appears by default for the second performance measure and then run the tool (with that field blank)? Looking at the code, this might do what you want, or it might just default to using fpr instead. If it’s the latter, we need to tweak the code to interpret blank as “none”.

 

Best,

Jason

 

 

Hi,

First of all thank you for the brilliant MGET tool, which makes life for someone like me who struggles with R so much easier. I have, however, a question concerning binary classification options when using Random Forest: In the Tool Help, it is stated that I can either plot two selected performance measures against each other or just one (on the Y axis) against the model cutoff on the X axis. I am currently attempting the latter and I am struggling to understand how I specify that I do not want to select a second performance measure. As a default, there are always two measures (tpr, fpr) selected and while I am able to select other measures it’s not possible to select ‘cutoff’ or leave the second performance measure blank. So, what would I have to do to plot for example sensitivity against cutoff?

Thanks a lot for your help.

Regards,

Markus

_______________________________

Dr. Markus Diesing

Norges geologiske undersøkelse (NGU)

Geological Survey of Norway

phone: +47 7390 4309
e-mail:

web: www.ngu.no