[Met_help] [rt.rap.ucar.edu #59040] History for Interest funtion

[Randy Bullock via RT]

----------------------------------------------------------------
  Initial Request
----------------------------------------------------------------


Hi,

I would like to have more information about how the information threshold is computed in the MODE tool.

I have looked to the PDF documentation, but I still wonder what equation it gives at the end with the weight that I give to he different variables.


Also, in the documentation there is approximately the same explanation for every variable. Which looks like this :


centroid_dist = (
( 0.0, 1.0 )
( 60.0/grid_res, 1.0 )
( 600.0/grid_res, 0.0 )
);

Does this mean that there is three options. Can you explain this more thoroughly?


Thanks
Anna-Belle Filion 
 		 	   		  

----------------------------------------------------------------
  Complete Ticket History
----------------------------------------------------------------

Subject: Interest funtion
From: Randy Bullock
Time: Fri Nov 02 12:41:53 2012


Hello -

Thanks for your questions about MODE.  Since it's
hard to show function graphs and equations in plain
ascii text, I've attached a couple of images from
my MET tutorial talk on MODE.  Hopefully they will
illustrate the points I'm trying to make.

(1)  The final equation that MODE uses to calculate the
     total interest of an object pair is shown on the attached
     image obj_18_t.png.  There, T is the total interest, C_i
     is the confidence map associated to attribute # i,
     I_i is the interest map associated with that attrubute,
     and w_i is the weight associated with that attribute.
     alpha is a vector (ordered n-tuple) of all the attributes,
     and the sums are over all attributes.

     Basically this says that the total interest is a weighted
     average of the interest values for all the attributes.

(2)  Piecewise linear interest maps for attributes are specified in
     the config file by giving the coordinates of of the corners
     of the graph.  So, taking the centroid distance interest
     map for example:

       centroid_dist = {

         (0.0), 1.0)
         (60.0/grid_res, 1.0)
         (600.0/grid_res, 0.0)

      };

      Let's say that grid_res is 10 km.  This isn't very realistic,
      but it makes the arithmetic easy.  Then the above definition
      says that the graph of the centroid distance interest map
      consists of two straight lines sequentially joining the three
      points (0.0, 1.0), (6.0, 1.0) and (60.0, 1.0) in the x-y plane.
      (See the example graph in the attached image obj_16_t.png)

Hope this helps.  Let me know if you need more information.

Randy Bullock


On Wed, Oct 31, 2012 at 09:46:41AM -0600, John Halley Gotway via RT
wrote:
>
> Wed Oct 31 09:46:41 2012: Request 59040 was acted upon.
> Transaction: Given to bullock by johnhg
>        Queue: met_help
>      Subject: Interest funtion
>        Owner: bullock
>   Requestors: treblecharger34 at hotmail.com
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=59040 >
>
>
> This transaction appears to have no content

------------------------------------------------