[Met_help] [rt.rap.ucar.edu #86554] History for METviewer 2.6 issues?

John Halley Gotway via RT met_help at ucar.edu
Tue Jul 9 12:07:29 MDT 2019


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

Hi,

I had been making reliability diagrams using v2.5 without issue, but with
the upgrade to v2.6 I'm getting this warning:

==== Start Rscript error ==== Warning message: In if (is.na(listStat)) { :
the condition has length > 1 and only the first element will be used ====
End Rscript error ====

The plot generally looks normal except it has a gap in the middle of the
curves (at 0.45 on x-axis).

Does this change in behavior make sense based on the v2.6 changes - maybe
the inclusion of confidence intervals?  Is there something I can try
changing to avoid this issue?

ROC curves give me the same warning and no data plotted, but I have less
experience making that plot type.

Thanks,

Matt


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

Subject: METviewer 2.6 issues?
From: John Halley Gotway
Time: Fri Aug 10 12:39:58 2018

Matt,

Thanks for letting us know about this change in behavior in mv2.6.
Tatiana
looked into it and found that the hole was caused by no values falling
in
that particular probability bin.  Previous versions of METviewer
connected
the reliability line across that NA value, but mv2.6 plotted it as a
"hole".  Tatiana just fixed it and redeployed mv2.6.  Please try
running
again and confirm that:
   (1) The warning message is gone and (2) the lines of the
reliability
diagram now connect across that NA value.

mv2.6 includes some code consolidation and cleanup and this small
change
was likely an unintended consequence of that.  Fortunately, it was
very
easy to fix.

Please let us know if you have other questions or encounter any
unexpected
behavior.

Thanks,
John


On Fri, Aug 10, 2018 at 11:29 AM matthew.pyle at noaa.gov via RT <
met_help at ucar.edu> wrote:

>
> Fri Aug 10 11:28:37 2018: Request 86554 was acted upon.
> Transaction: Ticket created by Matthew.Pyle at noaa.gov
>        Queue: met_help
>      Subject: METviewer 2.6 issues?
>        Owner: Nobody
>   Requestors: Matthew.Pyle at noaa.gov
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=86554 >
>
>
> Hi,
>
> I had been making reliability diagrams using v2.5 without issue, but
with
> the upgrade to v2.6 I'm getting this warning:
>
> ==== Start Rscript error ==== Warning message: In if
(is.na(listStat)) { :
> the condition has length > 1 and only the first element will be used
====
> End Rscript error ====
>
> The plot generally looks normal except it has a gap in the middle of
the
> curves (at 0.45 on x-axis).
>
> Does this change in behavior make sense based on the v2.6 changes -
maybe
> the inclusion of confidence intervals?  Is there something I can try
> changing to avoid this issue?
>
> ROC curves give me the same warning and no data plotted, but I have
less
> experience making that plot type.
>
> Thanks,
>
> Matt
>
>

------------------------------------------------
Subject: METviewer 2.6 issues?
From: "matthew.pyle at noaa.gov"
Time: Fri Aug 10 13:03:54 2018

Hi John/Tatiana,

Thanks for the quick response.  It does eliminate the warning message
and
the gap with reliability , BUT...

now it seems to be unable to plot multiple models simultaneously.
Plotting
each model individually looks fine (and fills in the gap as needed).
I'm
not seeing any warnings/errors why only one model is actually plotted.



On Fri, Aug 10, 2018 at 2:39 PM, John Halley Gotway via RT <
met_help at ucar.edu> wrote:

> Matt,
>
> Thanks for letting us know about this change in behavior in mv2.6.
Tatiana
> looked into it and found that the hole was caused by no values
falling in
> that particular probability bin.  Previous versions of METviewer
connected
> the reliability line across that NA value, but mv2.6 plotted it as a
> "hole".  Tatiana just fixed it and redeployed mv2.6.  Please try
running
> again and confirm that:
>    (1) The warning message is gone and (2) the lines of the
reliability
> diagram now connect across that NA value.
>
> mv2.6 includes some code consolidation and cleanup and this small
change
> was likely an unintended consequence of that.  Fortunately, it was
very
> easy to fix.
>
> Please let us know if you have other questions or encounter any
unexpected
> behavior.
>
> Thanks,
> John
>
>
> On Fri, Aug 10, 2018 at 11:29 AM matthew.pyle at noaa.gov via RT <
> met_help at ucar.edu> wrote:
>
> >
> > Fri Aug 10 11:28:37 2018: Request 86554 was acted upon.
> > Transaction: Ticket created by Matthew.Pyle at noaa.gov
> >        Queue: met_help
> >      Subject: METviewer 2.6 issues?
> >        Owner: Nobody
> >   Requestors: Matthew.Pyle at noaa.gov
> >       Status: new
> >  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=86554 >
> >
> >
> > Hi,
> >
> > I had been making reliability diagrams using v2.5 without issue,
but with
> > the upgrade to v2.6 I'm getting this warning:
> >
> > ==== Start Rscript error ==== Warning message: In if
(is.na(listStat))
> { :
> > the condition has length > 1 and only the first element will be
used ====
> > End Rscript error ====
> >
> > The plot generally looks normal except it has a gap in the middle
of the
> > curves (at 0.45 on x-axis).
> >
> > Does this change in behavior make sense based on the v2.6 changes
- maybe
> > the inclusion of confidence intervals?  Is there something I can
try
> > changing to avoid this issue?
> >
> > ROC curves give me the same warning and no data plotted, but I
have less
> > experience making that plot type.
> >
> > Thanks,
> >
> > Matt
> >
> >
>
>

------------------------------------------------
Subject: METviewer 2.6 issues?
From: Tatiana Burek
Time: Fri Aug 10 13:20:25 2018

Mat,
I apply another fix that should allow you to plot multiple models
simultaneously.

Tatiana
On Fri Aug 10 13:03:54 2018, Matthew.Pyle at noaa.gov wrote:
> Hi John/Tatiana,
>
> Thanks for the quick response.  It does eliminate the warning
message
> and
> the gap with reliability , BUT...
>
> now it seems to be unable to plot multiple models simultaneously.
> Plotting
> each model individually looks fine (and fills in the gap as needed).
> I'm
> not seeing any warnings/errors why only one model is actually
plotted.
>
>
>
> On Fri, Aug 10, 2018 at 2:39 PM, John Halley Gotway via RT <
> met_help at ucar.edu> wrote:
>
> > Matt,
> >
> > Thanks for letting us know about this change in behavior in mv2.6.
> > Tatiana
> > looked into it and found that the hole was caused by no values
> > falling in
> > that particular probability bin.  Previous versions of METviewer
> > connected
> > the reliability line across that NA value, but mv2.6 plotted it as
a
> > "hole".  Tatiana just fixed it and redeployed mv2.6.  Please try
> > running
> > again and confirm that:
> >    (1) The warning message is gone and (2) the lines of the
> > reliability
> > diagram now connect across that NA value.
> >
> > mv2.6 includes some code consolidation and cleanup and this small
> > change
> > was likely an unintended consequence of that.  Fortunately, it was
> > very
> > easy to fix.
> >
> > Please let us know if you have other questions or encounter any
> > unexpected
> > behavior.
> >
> > Thanks,
> > John
> >
> >
> > On Fri, Aug 10, 2018 at 11:29 AM matthew.pyle at noaa.gov via RT <
> > met_help at ucar.edu> wrote:
> >
> > >
> > > Fri Aug 10 11:28:37 2018: Request 86554 was acted upon.
> > > Transaction: Ticket created by Matthew.Pyle at noaa.gov
> > >        Queue: met_help
> > >      Subject: METviewer 2.6 issues?
> > >        Owner: Nobody
> > >   Requestors: Matthew.Pyle at noaa.gov
> > >       Status: new
> > >  Ticket <URL:
> > > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=86554 >
> > >
> > >
> > > Hi,
> > >
> > > I had been making reliability diagrams using v2.5 without issue,
> > > but with
> > > the upgrade to v2.6 I'm getting this warning:
> > >
> > > ==== Start Rscript error ==== Warning message: In if
> > > (is.na(listStat))
> > { :
> > > the condition has length > 1 and only the first element will be
> > > used ====
> > > End Rscript error ====
> > >
> > > The plot generally looks normal except it has a gap in the
middle
> > > of the
> > > curves (at 0.45 on x-axis).
> > >
> > > Does this change in behavior make sense based on the v2.6
changes -
> > > maybe
> > > the inclusion of confidence intervals?  Is there something I can
> > > try
> > > changing to avoid this issue?
> > >
> > > ROC curves give me the same warning and no data plotted, but I
have
> > > less
> > > experience making that plot type.
> > >
> > > Thanks,
> > >
> > > Matt
> > >
> > >
> >
> >



------------------------------------------------
Subject: METviewer 2.6 issues?
From: "matthew.pyle at noaa.gov"
Time: Fri Aug 10 14:04:00 2018

Hi Tatiana,

Thanks for that quick fix.  I can now plot reliabiliity diagrams for
multiple models simultaneously.

-Matt

On Fri, Aug 10, 2018 at 3:20 PM, Tatiana Burek via RT
<met_help at ucar.edu>
wrote:

> Mat,
> I apply another fix that should allow you to plot multiple models
> simultaneously.
>
> Tatiana
> On Fri Aug 10 13:03:54 2018, Matthew.Pyle at noaa.gov wrote:
> > Hi John/Tatiana,
> >
> > Thanks for the quick response.  It does eliminate the warning
message
> > and
> > the gap with reliability , BUT...
> >
> > now it seems to be unable to plot multiple models simultaneously.
> > Plotting
> > each model individually looks fine (and fills in the gap as
needed).
> > I'm
> > not seeing any warnings/errors why only one model is actually
plotted.
> >
> >
> >
> > On Fri, Aug 10, 2018 at 2:39 PM, John Halley Gotway via RT <
> > met_help at ucar.edu> wrote:
> >
> > > Matt,
> > >
> > > Thanks for letting us know about this change in behavior in
mv2.6.
> > > Tatiana
> > > looked into it and found that the hole was caused by no values
> > > falling in
> > > that particular probability bin.  Previous versions of METviewer
> > > connected
> > > the reliability line across that NA value, but mv2.6 plotted it
as a
> > > "hole".  Tatiana just fixed it and redeployed mv2.6.  Please try
> > > running
> > > again and confirm that:
> > >    (1) The warning message is gone and (2) the lines of the
> > > reliability
> > > diagram now connect across that NA value.
> > >
> > > mv2.6 includes some code consolidation and cleanup and this
small
> > > change
> > > was likely an unintended consequence of that.  Fortunately, it
was
> > > very
> > > easy to fix.
> > >
> > > Please let us know if you have other questions or encounter any
> > > unexpected
> > > behavior.
> > >
> > > Thanks,
> > > John
> > >
> > >
> > > On Fri, Aug 10, 2018 at 11:29 AM matthew.pyle at noaa.gov via RT <
> > > met_help at ucar.edu> wrote:
> > >
> > > >
> > > > Fri Aug 10 11:28:37 2018: Request 86554 was acted upon.
> > > > Transaction: Ticket created by Matthew.Pyle at noaa.gov
> > > >        Queue: met_help
> > > >      Subject: METviewer 2.6 issues?
> > > >        Owner: Nobody
> > > >   Requestors: Matthew.Pyle at noaa.gov
> > > >       Status: new
> > > >  Ticket <URL:
> > > > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=86554 >
> > > >
> > > >
> > > > Hi,
> > > >
> > > > I had been making reliability diagrams using v2.5 without
issue,
> > > > but with
> > > > the upgrade to v2.6 I'm getting this warning:
> > > >
> > > > ==== Start Rscript error ==== Warning message: In if
> > > > (is.na(listStat))
> > > { :
> > > > the condition has length > 1 and only the first element will
be
> > > > used ====
> > > > End Rscript error ====
> > > >
> > > > The plot generally looks normal except it has a gap in the
middle
> > > > of the
> > > > curves (at 0.45 on x-axis).
> > > >
> > > > Does this change in behavior make sense based on the v2.6
changes -
> > > > maybe
> > > > the inclusion of confidence intervals?  Is there something I
can
> > > > try
> > > > changing to avoid this issue?
> > > >
> > > > ROC curves give me the same warning and no data plotted, but I
have
> > > > less
> > > > experience making that plot type.
> > > >
> > > > Thanks,
> > > >
> > > > Matt
> > > >
> > > >
> > >
> > >
>
>
>
>

------------------------------------------------
Subject: METviewer 2.6 issues?
From: "matthew.pyle at noaa.gov"
Time: Wed Aug 15 07:44:15 2018

Have more changes been made to the v2.6 metviewer?  I returned to it
today,
and again am having issues doing multiple models.  A possible
complication
in what I'm trying to do is that the FCST_VAR definitions for the two
models I'm comparing are different as the probabilities are being
generated
from different codes:

APCP_03_ENS_FREQ_ge6.35
PROB(APCP_03>6.350)

This approach worked in the past, and again with v2.6 as of August
10th,
but now I'm getting this new error:

failed to run plot plot_20180815_133841 - reason: R error: ==== Start
Rscript error ==== Error in t.star[r, ] <- res[[r]] : number of items
to
replace is not a multiple of replacement length Error: $ operator is
invalid for atomic vectors Execution halted ==== End Rscript error
====
Populating /var/lib/tomcat6/webapps/metviewer/R_tmpl/rely.R_tmpl
/common/data/web/metviewer/dev/scripts/plot_20180815_133841.R Running
'/usr/bin/Rscript
/common/data/web/metviewer/dev/scripts/plot_20180815_133841.R' ====
Start
Rscript error ==== Error in file(file, "rt") : cannot open the
connection
Calls: read.delim -> read.table -> file In addition: Warning message:
In
file(file, "rt") : cannot open file
'/common/data/web/metviewer/dev/data/plot_20180815_133841.data': No
such
file or directory Execution halted ==== End Rscript error ====

Thanks,

Matt

On Fri, Aug 10, 2018 at 4:03 PM, Matthew Pyle - NOAA Federal <
matthew.pyle at noaa.gov> wrote:

> Hi Tatiana,
>
> Thanks for that quick fix.  I can now plot reliabiliity diagrams for
> multiple models simultaneously.
>
> -Matt
>
> On Fri, Aug 10, 2018 at 3:20 PM, Tatiana Burek via RT
<met_help at ucar.edu>
> wrote:
>
>> Mat,
>> I apply another fix that should allow you to plot multiple models
>> simultaneously.
>>
>> Tatiana
>> On Fri Aug 10 13:03:54 2018, Matthew.Pyle at noaa.gov wrote:
>> > Hi John/Tatiana,
>> >
>> > Thanks for the quick response.  It does eliminate the warning
message
>> > and
>> > the gap with reliability , BUT...
>> >
>> > now it seems to be unable to plot multiple models simultaneously.
>> > Plotting
>> > each model individually looks fine (and fills in the gap as
needed).
>> > I'm
>> > not seeing any warnings/errors why only one model is actually
plotted.
>> >
>> >
>> >
>> > On Fri, Aug 10, 2018 at 2:39 PM, John Halley Gotway via RT <
>> > met_help at ucar.edu> wrote:
>> >
>> > > Matt,
>> > >
>> > > Thanks for letting us know about this change in behavior in
mv2.6.
>> > > Tatiana
>> > > looked into it and found that the hole was caused by no values
>> > > falling in
>> > > that particular probability bin.  Previous versions of
METviewer
>> > > connected
>> > > the reliability line across that NA value, but mv2.6 plotted it
as a
>> > > "hole".  Tatiana just fixed it and redeployed mv2.6.  Please
try
>> > > running
>> > > again and confirm that:
>> > >    (1) The warning message is gone and (2) the lines of the
>> > > reliability
>> > > diagram now connect across that NA value.
>> > >
>> > > mv2.6 includes some code consolidation and cleanup and this
small
>> > > change
>> > > was likely an unintended consequence of that.  Fortunately, it
was
>> > > very
>> > > easy to fix.
>> > >
>> > > Please let us know if you have other questions or encounter any
>> > > unexpected
>> > > behavior.
>> > >
>> > > Thanks,
>> > > John
>> > >
>> > >
>> > > On Fri, Aug 10, 2018 at 11:29 AM matthew.pyle at noaa.gov via RT <
>> > > met_help at ucar.edu> wrote:
>> > >
>> > > >
>> > > > Fri Aug 10 11:28:37 2018: Request 86554 was acted upon.
>> > > > Transaction: Ticket created by Matthew.Pyle at noaa.gov
>> > > >        Queue: met_help
>> > > >      Subject: METviewer 2.6 issues?
>> > > >        Owner: Nobody
>> > > >   Requestors: Matthew.Pyle at noaa.gov
>> > > >       Status: new
>> > > >  Ticket <URL:
>> > > > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=86554 >
>> > > >
>> > > >
>> > > > Hi,
>> > > >
>> > > > I had been making reliability diagrams using v2.5 without
issue,
>> > > > but with
>> > > > the upgrade to v2.6 I'm getting this warning:
>> > > >
>> > > > ==== Start Rscript error ==== Warning message: In if
>> > > > (is.na(listStat))
>> > > { :
>> > > > the condition has length > 1 and only the first element will
be
>> > > > used ====
>> > > > End Rscript error ====
>> > > >
>> > > > The plot generally looks normal except it has a gap in the
middle
>> > > > of the
>> > > > curves (at 0.45 on x-axis).
>> > > >
>> > > > Does this change in behavior make sense based on the v2.6
changes -
>> > > > maybe
>> > > > the inclusion of confidence intervals?  Is there something I
can
>> > > > try
>> > > > changing to avoid this issue?
>> > > >
>> > > > ROC curves give me the same warning and no data plotted, but
I have
>> > > > less
>> > > > experience making that plot type.
>> > > >
>> > > > Thanks,
>> > > >
>> > > > Matt
>> > > >
>> > > >
>> > >
>> > >
>>
>>
>>
>>
>

------------------------------------------------
Subject: METviewer 2.6 issues?
From: Tatiana Burek
Time: Thu Aug 16 09:59:45 2018

Matt,
I apply another fix that should take care the problem you discovered.
We moved Rely stats calculations to a different script so it is
possible that we did not anticipate all use cases . This one was :
different number of records in each series for  the particular thresh.
Let me know if it works for you now.

Tatiana

On Wed Aug 15 07:44:15 2018, Matthew.Pyle at noaa.gov wrote:
> Have more changes been made to the v2.6 metviewer?  I returned to it
today,
> and again am having issues doing multiple models.  A possible
complication
> in what I'm trying to do is that the FCST_VAR definitions for the
two
> models I'm comparing are different as the probabilities are being
generated
> from different codes:
>
> APCP_03_ENS_FREQ_ge6.35
> PROB(APCP_03>6.350)
>
> This approach worked in the past, and again with v2.6 as of August
10th,
> but now I'm getting this new error:
>
> failed to run plot plot_20180815_133841 - reason: R error: ====
Start
> Rscript error ==== Error in t.star[r, ] <- res[[r]] : number of
items to
> replace is not a multiple of replacement length Error: $ operator is
> invalid for atomic vectors Execution halted ==== End Rscript error
====
> Populating /var/lib/tomcat6/webapps/metviewer/R_tmpl/rely.R_tmpl
> /common/data/web/metviewer/dev/scripts/plot_20180815_133841.R
Running
> '/usr/bin/Rscript
> /common/data/web/metviewer/dev/scripts/plot_20180815_133841.R' ====
Start
> Rscript error ==== Error in file(file, "rt") : cannot open the
connection
> Calls: read.delim -> read.table -> file In addition: Warning
message: In
> file(file, "rt") : cannot open file
> '/common/data/web/metviewer/dev/data/plot_20180815_133841.data': No
such
> file or directory Execution halted ==== End Rscript error ====
>
> Thanks,
>
> Matt
>
> On Fri, Aug 10, 2018 at 4:03 PM, Matthew Pyle - NOAA Federal <
> matthew.pyle at noaa.gov> wrote:
>
> > Hi Tatiana,
> >
> > Thanks for that quick fix.  I can now plot reliabiliity diagrams
for
> > multiple models simultaneously.
> >
> > -Matt
> >
> > On Fri, Aug 10, 2018 at 3:20 PM, Tatiana Burek via RT
<met_help at ucar.edu>
> > wrote:
> >
> >> Mat,
> >> I apply another fix that should allow you to plot multiple models
> >> simultaneously.
> >>
> >> Tatiana
> >> On Fri Aug 10 13:03:54 2018, Matthew.Pyle at noaa.gov wrote:
> >> > Hi John/Tatiana,
> >> >
> >> > Thanks for the quick response.  It does eliminate the warning
message
> >> > and
> >> > the gap with reliability , BUT...
> >> >
> >> > now it seems to be unable to plot multiple models
simultaneously.
> >> > Plotting
> >> > each model individually looks fine (and fills in the gap as
needed).
> >> > I'm
> >> > not seeing any warnings/errors why only one model is actually
plotted.
> >> >
> >> >
> >> >
> >> > On Fri, Aug 10, 2018 at 2:39 PM, John Halley Gotway via RT <
> >> > met_help at ucar.edu> wrote:
> >> >
> >> > > Matt,
> >> > >
> >> > > Thanks for letting us know about this change in behavior in
mv2.6.
> >> > > Tatiana
> >> > > looked into it and found that the hole was caused by no
values
> >> > > falling in
> >> > > that particular probability bin.  Previous versions of
METviewer
> >> > > connected
> >> > > the reliability line across that NA value, but mv2.6 plotted
it as a
> >> > > "hole".  Tatiana just fixed it and redeployed mv2.6.  Please
try
> >> > > running
> >> > > again and confirm that:
> >> > >    (1) The warning message is gone and (2) the lines of the
> >> > > reliability
> >> > > diagram now connect across that NA value.
> >> > >
> >> > > mv2.6 includes some code consolidation and cleanup and this
small
> >> > > change
> >> > > was likely an unintended consequence of that.  Fortunately,
it was
> >> > > very
> >> > > easy to fix.
> >> > >
> >> > > Please let us know if you have other questions or encounter
any
> >> > > unexpected
> >> > > behavior.
> >> > >
> >> > > Thanks,
> >> > > John
> >> > >
> >> > >
> >> > > On Fri, Aug 10, 2018 at 11:29 AM matthew.pyle at noaa.gov via RT
<
> >> > > met_help at ucar.edu> wrote:
> >> > >
> >> > > >
> >> > > > Fri Aug 10 11:28:37 2018: Request 86554 was acted upon.
> >> > > > Transaction: Ticket created by Matthew.Pyle at noaa.gov
> >> > > >        Queue: met_help
> >> > > >      Subject: METviewer 2.6 issues?
> >> > > >        Owner: Nobody
> >> > > >   Requestors: Matthew.Pyle at noaa.gov
> >> > > >       Status: new
> >> > > >  Ticket <URL:
> >> > > > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=86554 >
> >> > > >
> >> > > >
> >> > > > Hi,
> >> > > >
> >> > > > I had been making reliability diagrams using v2.5 without
issue,
> >> > > > but with
> >> > > > the upgrade to v2.6 I'm getting this warning:
> >> > > >
> >> > > > ==== Start Rscript error ==== Warning message: In if
> >> > > > (is.na(listStat))
> >> > > { :
> >> > > > the condition has length > 1 and only the first element
will be
> >> > > > used ====
> >> > > > End Rscript error ====
> >> > > >
> >> > > > The plot generally looks normal except it has a gap in the
middle
> >> > > > of the
> >> > > > curves (at 0.45 on x-axis).
> >> > > >
> >> > > > Does this change in behavior make sense based on the v2.6
changes -
> >> > > > maybe
> >> > > > the inclusion of confidence intervals?  Is there something
I can
> >> > > > try
> >> > > > changing to avoid this issue?
> >> > > >
> >> > > > ROC curves give me the same warning and no data plotted,
but I have
> >> > > > less
> >> > > > experience making that plot type.
> >> > > >
> >> > > > Thanks,
> >> > > >
> >> > > > Matt
> >> > > >
> >> > > >
> >> > >
> >> > >
> >>
> >>
> >>
> >>
> >



------------------------------------------------
Subject: METviewer 2.6 issues?
From: "matthew.pyle at noaa.gov"
Time: Thu Aug 16 10:22:58 2018

Thanks - it is working well again for me!

-Matt

On Thu, Aug 16, 2018 at 11:59 AM, Tatiana Burek via RT
<met_help at ucar.edu>
wrote:

> Matt,
> I apply another fix that should take care the problem you
discovered.
> We moved Rely stats calculations to a different script so it is
possible
> that we did not anticipate all use cases . This one was : different
number
> of records in each series for  the particular thresh.
> Let me know if it works for you now.
>
> Tatiana
>
> On Wed Aug 15 07:44:15 2018, Matthew.Pyle at noaa.gov wrote:
> > Have more changes been made to the v2.6 metviewer?  I returned to
it
> today,
> > and again am having issues doing multiple models.  A possible
> complication
> > in what I'm trying to do is that the FCST_VAR definitions for the
two
> > models I'm comparing are different as the probabilities are being
> generated
> > from different codes:
> >
> > APCP_03_ENS_FREQ_ge6.35
> > PROB(APCP_03>6.350)
> >
> > This approach worked in the past, and again with v2.6 as of August
10th,
> > but now I'm getting this new error:
> >
> > failed to run plot plot_20180815_133841 - reason: R error: ====
Start
> > Rscript error ==== Error in t.star[r, ] <- res[[r]] : number of
items to
> > replace is not a multiple of replacement length Error: $ operator
is
> > invalid for atomic vectors Execution halted ==== End Rscript error
====
> > Populating /var/lib/tomcat6/webapps/metviewer/R_tmpl/rely.R_tmpl
> > /common/data/web/metviewer/dev/scripts/plot_20180815_133841.R
Running
> > '/usr/bin/Rscript
> > /common/data/web/metviewer/dev/scripts/plot_20180815_133841.R'
====
> Start
> > Rscript error ==== Error in file(file, "rt") : cannot open the
connection
> > Calls: read.delim -> read.table -> file In addition: Warning
message: In
> > file(file, "rt") : cannot open file
> > '/common/data/web/metviewer/dev/data/plot_20180815_133841.data':
No such
> > file or directory Execution halted ==== End Rscript error ====
> >
> > Thanks,
> >
> > Matt
> >
> > On Fri, Aug 10, 2018 at 4:03 PM, Matthew Pyle - NOAA Federal <
> > matthew.pyle at noaa.gov> wrote:
> >
> > > Hi Tatiana,
> > >
> > > Thanks for that quick fix.  I can now plot reliabiliity diagrams
for
> > > multiple models simultaneously.
> > >
> > > -Matt
> > >
> > > On Fri, Aug 10, 2018 at 3:20 PM, Tatiana Burek via RT <
> met_help at ucar.edu>
> > > wrote:
> > >
> > >> Mat,
> > >> I apply another fix that should allow you to plot multiple
models
> > >> simultaneously.
> > >>
> > >> Tatiana
> > >> On Fri Aug 10 13:03:54 2018, Matthew.Pyle at noaa.gov wrote:
> > >> > Hi John/Tatiana,
> > >> >
> > >> > Thanks for the quick response.  It does eliminate the warning
> message
> > >> > and
> > >> > the gap with reliability , BUT...
> > >> >
> > >> > now it seems to be unable to plot multiple models
simultaneously.
> > >> > Plotting
> > >> > each model individually looks fine (and fills in the gap as
needed).
> > >> > I'm
> > >> > not seeing any warnings/errors why only one model is actually
> plotted.
> > >> >
> > >> >
> > >> >
> > >> > On Fri, Aug 10, 2018 at 2:39 PM, John Halley Gotway via RT <
> > >> > met_help at ucar.edu> wrote:
> > >> >
> > >> > > Matt,
> > >> > >
> > >> > > Thanks for letting us know about this change in behavior in
mv2.6.
> > >> > > Tatiana
> > >> > > looked into it and found that the hole was caused by no
values
> > >> > > falling in
> > >> > > that particular probability bin.  Previous versions of
METviewer
> > >> > > connected
> > >> > > the reliability line across that NA value, but mv2.6
plotted it
> as a
> > >> > > "hole".  Tatiana just fixed it and redeployed mv2.6.
Please try
> > >> > > running
> > >> > > again and confirm that:
> > >> > >    (1) The warning message is gone and (2) the lines of the
> > >> > > reliability
> > >> > > diagram now connect across that NA value.
> > >> > >
> > >> > > mv2.6 includes some code consolidation and cleanup and this
small
> > >> > > change
> > >> > > was likely an unintended consequence of that.  Fortunately,
it was
> > >> > > very
> > >> > > easy to fix.
> > >> > >
> > >> > > Please let us know if you have other questions or encounter
any
> > >> > > unexpected
> > >> > > behavior.
> > >> > >
> > >> > > Thanks,
> > >> > > John
> > >> > >
> > >> > >
> > >> > > On Fri, Aug 10, 2018 at 11:29 AM matthew.pyle at noaa.gov via
RT <
> > >> > > met_help at ucar.edu> wrote:
> > >> > >
> > >> > > >
> > >> > > > Fri Aug 10 11:28:37 2018: Request 86554 was acted upon.
> > >> > > > Transaction: Ticket created by Matthew.Pyle at noaa.gov
> > >> > > >        Queue: met_help
> > >> > > >      Subject: METviewer 2.6 issues?
> > >> > > >        Owner: Nobody
> > >> > > >   Requestors: Matthew.Pyle at noaa.gov
> > >> > > >       Status: new
> > >> > > >  Ticket <URL:
> > >> > > > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=86554 >
> > >> > > >
> > >> > > >
> > >> > > > Hi,
> > >> > > >
> > >> > > > I had been making reliability diagrams using v2.5 without
issue,
> > >> > > > but with
> > >> > > > the upgrade to v2.6 I'm getting this warning:
> > >> > > >
> > >> > > > ==== Start Rscript error ==== Warning message: In if
> > >> > > > (is.na(listStat))
> > >> > > { :
> > >> > > > the condition has length > 1 and only the first element
will be
> > >> > > > used ====
> > >> > > > End Rscript error ====
> > >> > > >
> > >> > > > The plot generally looks normal except it has a gap in
the
> middle
> > >> > > > of the
> > >> > > > curves (at 0.45 on x-axis).
> > >> > > >
> > >> > > > Does this change in behavior make sense based on the v2.6
> changes -
> > >> > > > maybe
> > >> > > > the inclusion of confidence intervals?  Is there
something I can
> > >> > > > try
> > >> > > > changing to avoid this issue?
> > >> > > >
> > >> > > > ROC curves give me the same warning and no data plotted,
but I
> have
> > >> > > > less
> > >> > > > experience making that plot type.
> > >> > > >
> > >> > > > Thanks,
> > >> > > >
> > >> > > > Matt
> > >> > > >
> > >> > > >
> > >> > >
> > >> > >
> > >>
> > >>
> > >>
> > >>
> > >
>
>
>
>

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


More information about the Met_help mailing list