[Met_help] [rt.rap.ucar.edu #70317] History for ticket 70133

John Halley Gotway via RT met_help at ucar.edu
Thu Jan 15 10:08:24 MST 2015


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

Hi John
this is query again reg the MCTC . 
You had explained me the 3x3 contingency table. 
I had selected4 threholds, say >25, >35, >45 and >55. So in this case I am getting 7 categories which are listed below. 

1. >25
2. B/w 25 and 35
3. >35
4. B/w 35 and 45
5. >45
6. b/w 45 and 55
7. >55. 

As an example, I have the following dataset, 
OBS                 FCST
25                     67
34                     10
50                     76
87                     60
100                   10
45                     25
34                     45 
29                     67
73                     79
56                     10
56                     23
78                      34
21                      89
32                       56

So These are 2 ques:
1. In the case of MCTC, we are NOT calculating the A/B/C?D as we do in the traditional Approach???.  will each category (7 in my case) will have individual A,B,C,D???
2. how will I get 5x5 contingency table given 4 categories???

Kindly explain. 
Geeta


 		 	   		  

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

Subject: ticket 70133
From: John Halley Gotway
Time: Fri Jan 09 13:46:45 2015

Geeta,

Using N thresholds defines N+1 categories.  So your 4 thresholds
define a
5x5 contingency table.  Those 5 categories are defined as:
  1. x < 25
  2. 25 <= x < 35
  3. 35 <= x < 45
  4. 45 <= x < 55
  5. x > 55

Each value should fall into exactly 1 of those 5 categories.

Make sense?

Thanks,
John

On Fri, Jan 9, 2015 at 10:31 AM, Geeta Geeta via RT
<met_help at ucar.edu>
wrote:

>
> Fri Jan 09 10:31:32 2015: Request 70317 was acted upon.
> Transaction: Ticket created by geeta124 at hotmail.com
>        Queue: met_help
>      Subject: ticket 70133
>        Owner: Nobody
>   Requestors: geeta124 at hotmail.com
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70317 >
>
>
> Hi John
> this is query again reg the MCTC .
> You had explained me the 3x3 contingency table.
> I had selected4 threholds, say >25, >35, >45 and >55. So in this
case I am
> getting 7 categories which are listed below.
>
> 1. >25
> 2. B/w 25 and 35
> 3. >35
> 4. B/w 35 and 45
> 5. >45
> 6. b/w 45 and 55
> 7. >55.
>
> As an example, I have the following dataset,
> OBS                 FCST
> 25                     67
> 34                     10
> 50                     76
> 87                     60
> 100                   10
> 45                     25
> 34                     45
> 29                     67
> 73                     79
> 56                     10
> 56                     23
> 78                      34
> 21                      89
> 32                       56
>
> So These are 2 ques:
> 1. In the case of MCTC, we are NOT calculating the A/B/C?D as we do
in the
> traditional Approach???.  will each category (7 in my case) will
have
> individual A,B,C,D???
> 2. how will I get 5x5 contingency table given 4 categories???
>
> Kindly explain.
> Geeta
>
>
>
>

------------------------------------------------
Subject: ticket 70133
From: Geeta Geeta
Time: Sat Jan 10 05:56:00 2015

Ok John. That's fine. Kindly answer a few of my doubts that I
expressed in the mail.
So each OBS/FCST pair, will A B C D be defined for each category??? as
in the thradioional Approach???
Geeta


> Subject: Re: [rt.rap.ucar.edu #70317] ticket 70133
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Fri, 9 Jan 2015 13:46:45 -0700
>
> Geeta,
>
> Using N thresholds defines N+1 categories.  So your 4 thresholds
define a
> 5x5 contingency table.  Those 5 categories are defined as:
>   1. x < 25
>   2. 25 <= x < 35
>   3. 35 <= x < 45
>   4. 45 <= x < 55
>   5. x > 55
>
> Each value should fall into exactly 1 of those 5 categories.
>
> Make sense?
>
> Thanks,
> John
>
> On Fri, Jan 9, 2015 at 10:31 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > Fri Jan 09 10:31:32 2015: Request 70317 was acted upon.
> > Transaction: Ticket created by geeta124 at hotmail.com
> >        Queue: met_help
> >      Subject: ticket 70133
> >        Owner: Nobody
> >   Requestors: geeta124 at hotmail.com
> >       Status: new
> >  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70317 >
> >
> >
> > Hi John
> > this is query again reg the MCTC .
> > You had explained me the 3x3 contingency table.
> > I had selected4 threholds, say >25, >35, >45 and >55. So in this
case I am
> > getting 7 categories which are listed below.
> >
> > 1. >25
> > 2. B/w 25 and 35
> > 3. >35
> > 4. B/w 35 and 45
> > 5. >45
> > 6. b/w 45 and 55
> > 7. >55.
> >
> > As an example, I have the following dataset,
> > OBS                 FCST
> > 25                     67
> > 34                     10
> > 50                     76
> > 87                     60
> > 100                   10
> > 45                     25
> > 34                     45
> > 29                     67
> > 73                     79
> > 56                     10
> > 56                     23
> > 78                      34
> > 21                      89
> > 32                       56
> >
> > So These are 2 ques:
> > 1. In the case of MCTC, we are NOT calculating the A/B/C?D as we
do in the
> > traditional Approach???.  will each category (7 in my case) will
have
> > individual A,B,C,D???
> > 2. how will I get 5x5 contingency table given 4 categories???
> >
> > Kindly explain.
> > Geeta
> >
> >
> >
> >
>

------------------------------------------------
Subject: ticket 70133
From: John Halley Gotway
Time: Mon Jan 12 14:39:42 2015

Geeta,

In the "traditional" approach, you use a singe threshold to define an
2x2
contingency table.  By way of example, suppose we have 2-meter
temperature
in Celcius and threshold it at freezing using >0.  Depending on
whether or
not the forecast and observation values are freezing, you would add
one to
either A, B, C, or D in the 2x2 contingency table.  But we could also
refer
to those table entries using a 0/1 for the forecast slot and a 0/1 for
the
observation slot.  A "hit" would go into the (1,1) table entry and a
correct negative would go into the (0,0) table entry.

Next, instead of using 1 threshold, let's use 3, like this: >0, >10,
>20.
Rather than having a yes/no, we now have 3 different categories for
the
forecast and observation values.  They could be less than 0, 0 to 10,
10 to
20, or greater than 20.  We figure out which category the forecast
goes
into (0, 1, 2, or 3) and which category the observation goes into (0,
1, 2,
3).  Since we have 4 possible categories, we have a 4x4 table.  For a
forecast of 15 and an observation of 30, the forecast is in category 2
and
the observation is in category 3.  So we'd increment table entry (2,
3).
We do that for all of the fcst/obs pairs of values.

Once we have the 4x4 table filled out, there are only a handful of
statistics we can generate from it, including accuracy.  Accuracy is
just
the percent of pairs where the forecast and observation fell into the
same
category.  So it's the percentage of the table that was on the
diagonal.

Hope that helps.

Thanks,
John



On Sat, Jan 10, 2015 at 5:56 AM, Geeta Geeta via RT
<met_help at ucar.edu>
wrote:

>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70317 >
>
> Ok John. That's fine. Kindly answer a few of my doubts that I
expressed in
> the mail.
> So each OBS/FCST pair, will A B C D be defined for each category???
as in
> the thradioional Approach???
> Geeta
>
>
> > Subject: Re: [rt.rap.ucar.edu #70317] ticket 70133
> > From: met_help at ucar.edu
> > To: geeta124 at hotmail.com
> > Date: Fri, 9 Jan 2015 13:46:45 -0700
> >
> > Geeta,
> >
> > Using N thresholds defines N+1 categories.  So your 4 thresholds
define a
> > 5x5 contingency table.  Those 5 categories are defined as:
> >   1. x < 25
> >   2. 25 <= x < 35
> >   3. 35 <= x < 45
> >   4. 45 <= x < 55
> >   5. x > 55
> >
> > Each value should fall into exactly 1 of those 5 categories.
> >
> > Make sense?
> >
> > Thanks,
> > John
> >
> > On Fri, Jan 9, 2015 at 10:31 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > wrote:
> >
> > >
> > > Fri Jan 09 10:31:32 2015: Request 70317 was acted upon.
> > > Transaction: Ticket created by geeta124 at hotmail.com
> > >        Queue: met_help
> > >      Subject: ticket 70133
> > >        Owner: Nobody
> > >   Requestors: geeta124 at hotmail.com
> > >       Status: new
> > >  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=70317
> >
> > >
> > >
> > > Hi John
> > > this is query again reg the MCTC .
> > > You had explained me the 3x3 contingency table.
> > > I had selected4 threholds, say >25, >35, >45 and >55. So in this
case
> I am
> > > getting 7 categories which are listed below.
> > >
> > > 1. >25
> > > 2. B/w 25 and 35
> > > 3. >35
> > > 4. B/w 35 and 45
> > > 5. >45
> > > 6. b/w 45 and 55
> > > 7. >55.
> > >
> > > As an example, I have the following dataset,
> > > OBS                 FCST
> > > 25                     67
> > > 34                     10
> > > 50                     76
> > > 87                     60
> > > 100                   10
> > > 45                     25
> > > 34                     45
> > > 29                     67
> > > 73                     79
> > > 56                     10
> > > 56                     23
> > > 78                      34
> > > 21                      89
> > > 32                       56
> > >
> > > So These are 2 ques:
> > > 1. In the case of MCTC, we are NOT calculating the A/B/C?D as we
do in
> the
> > > traditional Approach???.  will each category (7 in my case) will
have
> > > individual A,B,C,D???
> > > 2. how will I get 5x5 contingency table given 4 categories???
> > >
> > > Kindly explain.
> > > Geeta
> > >
> > >
> > >
> > >
> >
>
>

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


More information about the Met_help mailing list