[Met_help] [rt.rap.ucar.edu #59427] History for Question about Stat analysis
John Halley Gotway via RT
met_help at ucar.edu
Mon Feb 4 09:04:52 MST 2013
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Dear met-help,
I am working with Stat analysis and this is the command that works:
/u/charo/METv3.1/bin/stat_analysis
-lookin /u/charo/MET_results/POINTSTAT/noviembre/tmp
/u/charo/MET_results/POINTSTAT/septiembre/tmp
/u/charo/MET_results/POINTSTAT/octubre/tmp -v 2 -job aggregate_stat
-line_type MPR -out_fcst_thresh le208.8 -out_obs_thresh le208.8
-out_line_type CTC -vx_mask 87155 -fcst_var TMP -fcst_lev P150 -fcst_lead
360000 -dump_row out/MPR_job.stat > resMPR_CTCprueba1.txt
And the output is:
COL_NAME: TOTAL FY_OY FY_ON FN_OY FN_ON
CTC: 85 21 9 9 46
I need to specify also in the same command, -out_fcst_thresh gt200
-out_obs_thresh
gt200 and -fcst_lev P150-100, then I tried but it didn’t work.
I use -job aggregate_stat because in the Poinstat configuration file I
don´t switch on the CTC option.
Thanks in advance,
Gisela
P.S= By the way, I would like to know if you receive the Pdf showing my
work with MET.
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #59427] Question about Stat analysis
From: John Halley Gotway
Time: Thu Nov 29 08:15:37 2012
Gisela,
Yes, we received your poster. Thanks for sending that.
Unfortunately,
nobody in our immediate group speaks Spanish, so we weren't able to
read
much of it. But I'm glad MET was useful to your evaluation.
Regarding your question, I'm not sure what "didn't work about it". I
expect that changing the thresholds from le208.8 to gt200 should work
just
fine. However, changing the -fcst_lev from P100 to P150-100 would be
more
problematic. The -fcst_lev option tells STAT-Analysis to filter the
data
using the FCST_LEV column of the MET output. STAT-Analysis is just
doing
string matching for that column's values. It's not actually doing any
numerical comparisons of the level values themselves.
Your original STAT-Analysis job found 85 matching MPR lines for
-fcst_lev
P150 since there were 85 lines with that *exact string* in the
FCST_LEV
column of the MET output you passed to the tool.
Changing to -fcst_lev P150-100 will only produce matches if that exact
string appears in the FCST_LEV column of the MET output. And that
will
only be the case if you've configured Point-Stat to verify over the
range
of levels. Please check the .stat files from Point-Stat to see if it
contains that string in the FCST_LEV column. Point-Stat can be
configured
to very at a single level (P150) and/or over a range of levels (P150-
100).
Hopefully that's clear. If you have more specific questions, please
send
me some sample data, the version of MET you're running, the exact
command
you're running, and a description of the problem.
Thanks,
John Halley Gotway
>
> Wed Nov 28 09:01:26 2012: Request 59427 was acted upon.
> Transaction: Ticket created by charogiselad at gmail.com
> Queue: met_help
> Subject: Question about Stat analysis
> Owner: Nobody
> Requestors: charogiselad at gmail.com
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=59427 >
>
>
> Dear met-help,
>
> I am working with Stat analysis and this is the command that
works:
>
> /u/charo/METv3.1/bin/stat_analysis
>
> -lookin /u/charo/MET_results/POINTSTAT/noviembre/tmp
> /u/charo/MET_results/POINTSTAT/septiembre/tmp
> /u/charo/MET_results/POINTSTAT/octubre/tmp -v 2 -job aggregate_stat
> -line_type MPR -out_fcst_thresh le208.8 -out_obs_thresh le208.8
> -out_line_type CTC -vx_mask 87155 -fcst_var TMP -fcst_lev P150
-fcst_lead
> 360000 -dump_row out/MPR_job.stat > resMPR_CTCprueba1.txt
>
>
>
> And the output is:
>
> COL_NAME: TOTAL FY_OY FY_ON FN_OY FN_ON
>
> CTC: 85 21 9 9 46
>
>
>
> I need to specify also in the same command, -out_fcst_thresh gt200
> -out_obs_thresh
> gt200 and -fcst_lev P150-100, then I tried but it didn’t work.
>
> I use -job aggregate_stat because in the Poinstat configuration file
I
> don´t switch on the CTC option.
>
> Thanks in advance,
>
> Gisela
>
> P.S= By the way, I would like to know if you receive the Pdf
showing my
> work with MET.
>
------------------------------------------------
Subject: Question about Stat analysis
From: Gisela Daniela Charó
Time: Tue Dec 18 09:20:24 2012
Dear John,
Thanks for the answer, I understand your explanation about what
FCST_LEV
should be configured.
But I have another question:
I need to include (203<=threshold<=208)
-out_fcst_thresh le208 -out_fcst_thresh gt203
and it doesn't work, the same for -out_obs_thresh
Thanks in advance
Gisela
2012/11/29 John Halley Gotway via RT <met_help at ucar.edu>
> Gisela,
>
> Yes, we received your poster. Thanks for sending that.
Unfortunately,
> nobody in our immediate group speaks Spanish, so we weren't able to
read
> much of it. But I'm glad MET was useful to your evaluation.
>
> Regarding your question, I'm not sure what "didn't work about it".
I
> expect that changing the thresholds from le208.8 to gt200 should
work just
> fine. However, changing the -fcst_lev from P100 to P150-100 would
be more
> problematic. The -fcst_lev option tells STAT-Analysis to filter the
data
> using the FCST_LEV column of the MET output. STAT-Analysis is just
doing
> string matching for that column's values. It's not actually doing
any
> numerical comparisons of the level values themselves.
>
> Your original STAT-Analysis job found 85 matching MPR lines for
-fcst_lev
> P150 since there were 85 lines with that *exact string* in the
FCST_LEV
> column of the MET output you passed to the tool.
>
> Changing to -fcst_lev P150-100 will only produce matches if that
exact
> string appears in the FCST_LEV column of the MET output. And that
will
> only be the case if you've configured Point-Stat to verify over the
range
> of levels. Please check the .stat files from Point-Stat to see if
it
> contains that string in the FCST_LEV column. Point-Stat can be
configured
> to very at a single level (P150) and/or over a range of levels
(P150-100).
>
> Hopefully that's clear. If you have more specific questions, please
send
> me some sample data, the version of MET you're running, the exact
command
> you're running, and a description of the problem.
>
> Thanks,
> John Halley Gotway
>
> >
> > Wed Nov 28 09:01:26 2012: Request 59427 was acted upon.
> > Transaction: Ticket created by charogiselad at gmail.com
> > Queue: met_help
> > Subject: Question about Stat analysis
> > Owner: Nobody
> > Requestors: charogiselad at gmail.com
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=59427 >
> >
> >
> > Dear met-help,
> >
> > I am working with Stat analysis and this is the command that
works:
> >
> > /u/charo/METv3.1/bin/stat_analysis
> >
> > -lookin /u/charo/MET_results/POINTSTAT/noviembre/tmp
> > /u/charo/MET_results/POINTSTAT/septiembre/tmp
> > /u/charo/MET_results/POINTSTAT/octubre/tmp -v 2 -job
aggregate_stat
> > -line_type MPR -out_fcst_thresh le208.8 -out_obs_thresh le208.8
> > -out_line_type CTC -vx_mask 87155 -fcst_var TMP -fcst_lev P150
> -fcst_lead
> > 360000 -dump_row out/MPR_job.stat > resMPR_CTCprueba1.txt
> >
> >
> >
> > And the output is:
> >
> > COL_NAME: TOTAL FY_OY FY_ON FN_OY FN_ON
> >
> > CTC: 85 21 9 9 46
> >
> >
> >
> > I need to specify also in the same command, -out_fcst_thresh gt200
> > -out_obs_thresh
> > gt200 and -fcst_lev P150-100, then I tried but it didn’t work.
> >
> > I use -job aggregate_stat because in the Poinstat configuration
file I
> > don´t switch on the CTC option.
> >
> > Thanks in advance,
> >
> > Gisela
> >
> > P.S= By the way, I would like to know if you receive the Pdf
showing my
> > work with MET.
> >
>
>
>
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #59427] Question about Stat analysis
From: John Halley Gotway
Time: Tue Dec 18 10:00:14 2012
Gisela,
Hmmm, it sounds like rather than specifying a single threshold, you
want to specify a range of thresholds.
When you specify a single threshold like "<=208", you partition the
space into 2 pieces. Below the threshold is considered an event,
while above is considered a non-event. We can only use a single
threshold to define 2x2 contingency tables.
But you'd like to specify two thresholds like "203<=threshold<=208",
which would partition the space into 3 pieces. I assume you want an
event to be anything falling between those two thresholds, and
a non-event to be everything else. MET cannot define 2x2 contingency
table in this way.
However, MET does have support for multi-category contingency tables
in the MCTC and MCTS line types. You can specify multiple thresholds
to partition the forecast and observation values into an
arbitrary number of categories. And there are a few statistics that
can be derived from this nxn contingency table in the MCTS line type.
If you have questions about *why* MET doesn't support the type of 2x2
contingency table definition you're trying to do, I'd refer you to our
resident statistician, Tressa Fowler.
Hopefully that helps.
Thanks,
John
On 12/18/2012 09:20 AM, Gisela Daniela Charó via RT wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=59427 >
>
> Dear John,
> Thanks for the answer, I understand your explanation about what
FCST_LEV
> should be configured.
> But I have another question:
> I need to include (203<=threshold<=208)
> -out_fcst_thresh le208 -out_fcst_thresh gt203
> and it doesn't work, the same for -out_obs_thresh
>
> Thanks in advance
> Gisela
>
>
>
> 2012/11/29 John Halley Gotway via RT <met_help at ucar.edu>
>
>> Gisela,
>>
>> Yes, we received your poster. Thanks for sending that.
Unfortunately,
>> nobody in our immediate group speaks Spanish, so we weren't able to
read
>> much of it. But I'm glad MET was useful to your evaluation.
>>
>> Regarding your question, I'm not sure what "didn't work about it".
I
>> expect that changing the thresholds from le208.8 to gt200 should
work just
>> fine. However, changing the -fcst_lev from P100 to P150-100 would
be more
>> problematic. The -fcst_lev option tells STAT-Analysis to filter
the data
>> using the FCST_LEV column of the MET output. STAT-Analysis is just
doing
>> string matching for that column's values. It's not actually doing
any
>> numerical comparisons of the level values themselves.
>>
>> Your original STAT-Analysis job found 85 matching MPR lines for
-fcst_lev
>> P150 since there were 85 lines with that *exact string* in the
FCST_LEV
>> column of the MET output you passed to the tool.
>>
>> Changing to -fcst_lev P150-100 will only produce matches if that
exact
>> string appears in the FCST_LEV column of the MET output. And that
will
>> only be the case if you've configured Point-Stat to verify over the
range
>> of levels. Please check the .stat files from Point-Stat to see if
it
>> contains that string in the FCST_LEV column. Point-Stat can be
configured
>> to very at a single level (P150) and/or over a range of levels
(P150-100).
>>
>> Hopefully that's clear. If you have more specific questions,
please send
>> me some sample data, the version of MET you're running, the exact
command
>> you're running, and a description of the problem.
>>
>> Thanks,
>> John Halley Gotway
>>
>>>
>>> Wed Nov 28 09:01:26 2012: Request 59427 was acted upon.
>>> Transaction: Ticket created by charogiselad at gmail.com
>>> Queue: met_help
>>> Subject: Question about Stat analysis
>>> Owner: Nobody
>>> Requestors: charogiselad at gmail.com
>>> Status: new
>>> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=59427 >
>>>
>>>
>>> Dear met-help,
>>>
>>> I am working with Stat analysis and this is the command that
works:
>>>
>>> /u/charo/METv3.1/bin/stat_analysis
>>>
>>> -lookin /u/charo/MET_results/POINTSTAT/noviembre/tmp
>>> /u/charo/MET_results/POINTSTAT/septiembre/tmp
>>> /u/charo/MET_results/POINTSTAT/octubre/tmp -v 2 -job
aggregate_stat
>>> -line_type MPR -out_fcst_thresh le208.8 -out_obs_thresh le208.8
>>> -out_line_type CTC -vx_mask 87155 -fcst_var TMP -fcst_lev P150
>> -fcst_lead
>>> 360000 -dump_row out/MPR_job.stat > resMPR_CTCprueba1.txt
>>>
>>>
>>>
>>> And the output is:
>>>
>>> COL_NAME: TOTAL FY_OY FY_ON FN_OY FN_ON
>>>
>>> CTC: 85 21 9 9 46
>>>
>>>
>>>
>>> I need to specify also in the same command, -out_fcst_thresh gt200
>>> -out_obs_thresh
>>> gt200 and -fcst_lev P150-100, then I tried but it didn’t work.
>>>
>>> I use -job aggregate_stat because in the Poinstat configuration
file I
>>> don´t switch on the CTC option.
>>>
>>> Thanks in advance,
>>>
>>> Gisela
>>>
>>> P.S= By the way, I would like to know if you receive the Pdf
showing my
>>> work with MET.
>>>
>>
>>
>>
>>
------------------------------------------------
More information about the Met_help
mailing list