[Met_help] [rt.rap.ucar.edu #81713] History for Confidence Intervals for PSTD variables

John Halley Gotway via RT met_help at ucar.edu
Wed Jul 10 16:43:21 MDT 2019


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

Hi,

I'm running into some issues with plotting confidence intervals in
METViewer.  I've set bootstraping replications = 1000 and I chose "boot"
for the confidence interval type.  However, I am unable to plot any
confidence intervals for PSTD_ROC_AUC - stat_bcl and stat_bcu come up as NA
in the R data.

I am able to plot confidence intervals for the Brier Score, but when I try
to add a difference curve, I am unable to plot any confidence intervals for
that.  Here is part of the Rscript error: "zero-length arrow is of
indeterminate angle and so skipped"

I'm not really sure what I am doing incorrectly.  I've attached xml files
for my Brier Score and AUC score plots which use my mv_met_bblake_prob
database.

Thanks!
Ben


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

Subject: Confidence Intervals for PSTD variables
From: Tatiana Burek
Time: Wed Aug 23 13:45:50 2017

Ben,
METViewer doesn't support bootstrapping and CI for PSTD statistics (
PCT line type ).
I'm working on this now and will include into the next METViewer
release.

Tatiana

On Tue Aug 22 13:40:00 2017, benjamin.blake at noaa.gov wrote:
> Hi,
>
> I'm running into some issues with plotting confidence intervals in
> METViewer.  I've set bootstraping replications = 1000 and I chose
"boot"
> for the confidence interval type.  However, I am unable to plot any
> confidence intervals for PSTD_ROC_AUC - stat_bcl and stat_bcu come
up as NA
> in the R data.
>
> I am able to plot confidence intervals for the Brier Score, but when
I try
> to add a difference curve, I am unable to plot any confidence
intervals for
> that.  Here is part of the Rscript error: "zero-length arrow is of
> indeterminate angle and so skipped"
>
> I'm not really sure what I am doing incorrectly.  I've attached xml
files
> for my Brier Score and AUC score plots which use my
mv_met_bblake_prob
> database.
>
> Thanks!
> Ben



------------------------------------------------
Subject: Confidence Intervals for PSTD variables
From: Benjamin Blake - NOAA Affiliate
Time: Thu Aug 24 05:29:01 2017

Hi Tatiana,

Oh ok, that makes sense.  Thank you!

Ben

On Wed, Aug 23, 2017 at 3:45 PM, Tatiana Burek via RT
<met_help at ucar.edu>
wrote:

> Ben,
> METViewer doesn't support bootstrapping and CI for PSTD statistics (
PCT
> line type ).
> I'm working on this now and will include into the next METViewer
release.
>
> Tatiana
>
> On Tue Aug 22 13:40:00 2017, benjamin.blake at noaa.gov wrote:
> > Hi,
> >
> > I'm running into some issues with plotting confidence intervals in
> > METViewer.  I've set bootstraping replications = 1000 and I chose
"boot"
> > for the confidence interval type.  However, I am unable to plot
any
> > confidence intervals for PSTD_ROC_AUC - stat_bcl and stat_bcu come
up as
> NA
> > in the R data.
> >
> > I am able to plot confidence intervals for the Brier Score, but
when I
> try
> > to add a difference curve, I am unable to plot any confidence
intervals
> for
> > that.  Here is part of the Rscript error: "zero-length arrow is of
> > indeterminate angle and so skipped"
> >
> > I'm not really sure what I am doing incorrectly.  I've attached
xml files
> > for my Brier Score and AUC score plots which use my
mv_met_bblake_prob
> > database.
> >
> > Thanks!
> > Ben
>
>
>
>

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


More information about the Met_help mailing list