Some issues in reliability diagram and ranked histogram generation from METplotpy #2616
Replies: 15 comments 4 replies
-
I just ftp the sample files for testing |
Beta Was this translation helpful? Give feedback.
-
Thanks for the information. Due to resource constraints, we will not be able to fully address this issue until we resolve these constraints. In the meantime, I can download the sample data or have someone retrieve it if you can provide the host and directory. Regards, |
Beta Was this translation helpful? Give feedback.
-
Alicia,
Here is the response from DTC on the metplotpy discussion#2616
Binbin
…On Tue, Jun 18, 2024 at 2:15 PM bikegeek ***@***.***> wrote:
Thanks for the information. Due to resource constraints, we will not be
able to fully address this issue until we resolve these constraints. In the
meantime, I can download the sample data or have someone retrieve it if you
can provide the host and directory.
Regards,
Minna
—
Reply to this email directly, view it on GitHub
<#2616 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AQBMPFKNBQIKHBMAOOMZIHTZIB2KBAVCNFSM6AAAAABJOFZN3KVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TQMBZGUYDK>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
--
Binbin Zhou
Physical Scientist
Lynker at NOAA/NWS/NCEP/EMC
5830 University Research Ct.
College Park, MD 20740
***@***.***
301-683-3683
|
Beta Was this translation helpful? Give feedback.
-
I took a look at the data and reproduced your reliability diagram. It looks like the plot is treating your data as one dataset, rather than separating the points based on model. I'm in the process of figuring out how to set up the yaml config file to show each model as its own line. |
Beta Was this translation helpful? Give feedback.
-
I can create a plot with a single model, but haven't yet figured out how to
get *both* models as separate lines on the reliability diagram. I am also
investigating the mechanism of the inset histogram. I am not sure if it
was developed to create a single histogram with multiple-colored bars or
generate multiple inset histograms, or if it only supports one model.
…On Fri, Jul 12, 2024 at 9:07 AM BinbinZhou-NOAA ***@***.***> wrote:
Thanks for addressing this issue!
—
Reply to this email directly, view it on GitHub
<#2616 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AA4UJHVJ4BPYZXBRVSOLUB3ZL7WKHAVCNFSM6AAAAABJOFZN3KVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTAMBTGI4TCOI>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
Beta Was this translation helpful? Give feedback.
-
I'm not sure why there are double lines and text (No-Skill and No-resolution), as I don't observe it in the plot I generated with identical settings but different data: |
Beta Was this translation helpful? Give feedback.
-
Thanks for resolving this issue. Now I can duplicate the reliability plots for the two models as shown above. The issue for the 2 Binbin |
Beta Was this translation helpful? Give feedback.
-
Thanks for resolving this issue. Now I can duplicate the reliability plots
for the two models as shown above. The issue for the 2 No-Skill lines and 2
No-resolution lines in the first plot but not in the second plot, I guess
the reason is that the reliability curves in the second plot are pretty
close so that the 2 No-Skill lines and 2 No-resolution lines are overlapped
while the 2 reliability curves in the first plot are very different so
their No-skill and No-resolution lines are not overlapped. So the code
should plot the No-skill and No-resolution lines just once instead of 2
times for each model. As for the number of models, currently 2 models in
the same plot is our minimum requirement.
Binbin
…On Fri, Jul 12, 2024 at 7:22 PM bikegeek ***@***.***> wrote:
I'm not sure why there are double lines and text (No-Skill and
No-resolution), as I don't observe it in the plot I generated with
identical settings but different data:
plot_20240712_185518.png (view on web)
<https://github.com/user-attachments/assets/6dbfd5cc-25e1-4f4c-9e04-346ead2b3357>
—
Reply to this email directly, view it on GitHub
<#2616 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AQBMPFNQO6KLPAWNCKD3Q6DZMBQMNAVCNFSM6AAAAABJOFZN3KVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTAMBTGYZTANA>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
--
Binbin Zhou
Physical Scientist
Lynker at NOAA/NWS/NCEP/EMC
5830 University Research Ct.
College Park, MD 20740
***@***.***
301-683-3683
|
Beta Was this translation helpful? Give feedback.
-
One more important issue: the legend is still missing from the new plots
even though it is set to be True in
the ymal file:
show_legend:
- 'True'
- 'True'
Binbin
On Mon, Jul 15, 2024 at 9:58 AM Binbin Zhou - NOAA Affiliate <
***@***.***> wrote:
… Thanks for resolving this issue. Now I can duplicate the reliability plots
for the two models as shown above. The issue for the 2 No-Skill lines and 2
No-resolution lines in the first plot but not in the second plot, I guess
the reason is that the reliability curves in the second plot are pretty
close so that the 2 No-Skill lines and 2 No-resolution lines are overlapped
while the 2 reliability curves in the first plot are very different so
their No-skill and No-resolution lines are not overlapped. So the code
should plot the No-skill and No-resolution lines just once instead of 2
times for each model. As for the number of models, currently 2 models in
the same plot is our minimum requirement.
Binbin
On Fri, Jul 12, 2024 at 7:22 PM bikegeek ***@***.***> wrote:
> I'm not sure why there are double lines and text (No-Skill and
> No-resolution), as I don't observe it in the plot I generated with
> identical settings but different data:
> plot_20240712_185518.png (view on web)
> <https://github.com/user-attachments/assets/6dbfd5cc-25e1-4f4c-9e04-346ead2b3357>
>
> —
> Reply to this email directly, view it on GitHub
> <#2616 (comment)>,
> or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/AQBMPFNQO6KLPAWNCKD3Q6DZMBQMNAVCNFSM6AAAAABJOFZN3KVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTAMBTGYZTANA>
> .
> You are receiving this because you were mentioned.Message ID:
> ***@***.***>
>
--
Binbin Zhou
Physical Scientist
Lynker at NOAA/NWS/NCEP/EMC
5830 University Research Ct.
College Park, MD 20740
***@***.***
301-683-3683
--
Binbin Zhou
Physical Scientist
Lynker at NOAA/NWS/NCEP/EMC
5830 University Research Ct.
College Park, MD 20740
***@***.***
301-683-3683
|
Beta Was this translation helpful? Give feedback.
-
Good news, the reliability plot for 3 models (REFS_EAS, REFS_PROB and
HREF_PROB)
can be generated! But no legends are presented in the plot.
Binbin
[image: image.png]
On Mon, Jul 15, 2024 at 10:06 AM Binbin Zhou - NOAA Affiliate <
***@***.***> wrote:
… One more important issue: the legend is still missing from the new plots
even though it is set to be True in
the ymal file:
show_legend:
- 'True'
- 'True'
Binbin
On Mon, Jul 15, 2024 at 9:58 AM Binbin Zhou - NOAA Affiliate <
***@***.***> wrote:
> Thanks for resolving this issue. Now I can duplicate the reliability
> plots for the two models as shown above. The issue for the 2 No-Skill lines
> and 2 No-resolution lines in the first plot but not in the second plot, I
> guess the reason is that the reliability curves in the second plot are
> pretty close so that the 2 No-Skill lines and 2 No-resolution lines are
> overlapped while the 2 reliability curves in the first plot are very
> different so their No-skill and No-resolution lines are not overlapped. So
> the code should plot the No-skill and No-resolution lines just once instead
> of 2 times for each model. As for the number of models, currently 2 models
> in the same plot is our minimum requirement.
>
> Binbin
>
> On Fri, Jul 12, 2024 at 7:22 PM bikegeek ***@***.***>
> wrote:
>
>> I'm not sure why there are double lines and text (No-Skill and
>> No-resolution), as I don't observe it in the plot I generated with
>> identical settings but different data:
>> plot_20240712_185518.png (view on web)
>> <https://github.com/user-attachments/assets/6dbfd5cc-25e1-4f4c-9e04-346ead2b3357>
>>
>> —
>> Reply to this email directly, view it on GitHub
>> <#2616 (comment)>,
>> or unsubscribe
>> <https://github.com/notifications/unsubscribe-auth/AQBMPFNQO6KLPAWNCKD3Q6DZMBQMNAVCNFSM6AAAAABJOFZN3KVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTAMBTGYZTANA>
>> .
>> You are receiving this because you were mentioned.Message ID:
>> ***@***.***>
>>
>
>
> --
>
> Binbin Zhou
>
> Physical Scientist
>
> Lynker at NOAA/NWS/NCEP/EMC
>
> 5830 University Research Ct.
>
> College Park, MD 20740
>
> ***@***.***
>
> 301-683-3683
>
--
Binbin Zhou
Physical Scientist
Lynker at NOAA/NWS/NCEP/EMC
5830 University Research Ct.
College Park, MD 20740
***@***.***
301-683-3683
--
Binbin Zhou
Physical Scientist
Lynker at NOAA/NWS/NCEP/EMC
5830 University Research Ct.
College Park, MD 20740
***@***.***
301-683-3683
|
Beta Was this translation helpful? Give feedback.
-
I am looking into the legends. It looks like they are getting generated,
but I need to investigate why they aren't showing up on the plot.
On Mon, Jul 15, 2024 at 9:45 AM BinbinZhou-NOAA ***@***.***>
wrote:
… Good news, the reliability plot for 3 models (REFS_EAS, REFS_PROB and
HREF_PROB)
can be generated! But no legends are presented in the plot.
Binbin
[image: image.png]
On Mon, Jul 15, 2024 at 10:06 AM Binbin Zhou - NOAA Affiliate <
***@***.***> wrote:
> One more important issue: the legend is still missing from the new plots
> even though it is set to be True in
> the ymal file:
> show_legend:
> - 'True'
> - 'True'
>
> Binbin
>
> On Mon, Jul 15, 2024 at 9:58 AM Binbin Zhou - NOAA Affiliate <
> ***@***.***> wrote:
>
>> Thanks for resolving this issue. Now I can duplicate the reliability
>> plots for the two models as shown above. The issue for the 2 No-Skill
lines
>> and 2 No-resolution lines in the first plot but not in the second plot,
I
>> guess the reason is that the reliability curves in the second plot are
>> pretty close so that the 2 No-Skill lines and 2 No-resolution lines are
>> overlapped while the 2 reliability curves in the first plot are very
>> different so their No-skill and No-resolution lines are not overlapped.
So
>> the code should plot the No-skill and No-resolution lines just once
instead
>> of 2 times for each model. As for the number of models, currently 2
models
>> in the same plot is our minimum requirement.
>>
>> Binbin
>>
>> On Fri, Jul 12, 2024 at 7:22 PM bikegeek ***@***.***>
>> wrote:
>>
>>> I'm not sure why there are double lines and text (No-Skill and
>>> No-resolution), as I don't observe it in the plot I generated with
>>> identical settings but different data:
>>> plot_20240712_185518.png (view on web)
>>> <
https://github.com/user-attachments/assets/6dbfd5cc-25e1-4f4c-9e04-346ead2b3357>
>>>
>>> —
>>> Reply to this email directly, view it on GitHub
>>> <
#2616 (comment)>,
>>> or unsubscribe
>>> <
https://github.com/notifications/unsubscribe-auth/AQBMPFNQO6KLPAWNCKD3Q6DZMBQMNAVCNFSM6AAAAABJOFZN3KVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTAMBTGYZTANA>
>>> .
>>> You are receiving this because you were mentioned.Message ID:
>>> ***@***.***>
>>>
>>
>>
>> --
>>
>> Binbin Zhou
>>
>> Physical Scientist
>>
>> Lynker at NOAA/NWS/NCEP/EMC
>>
>> 5830 University Research Ct.
>>
>> College Park, MD 20740
>>
>> ***@***.***
>>
>> 301-683-3683
>>
>
>
> --
>
> Binbin Zhou
>
> Physical Scientist
>
> Lynker at NOAA/NWS/NCEP/EMC
>
> 5830 University Research Ct.
>
> College Park, MD 20740
>
> ***@***.***
>
> 301-683-3683
>
--
Binbin Zhou
Physical Scientist
Lynker at NOAA/NWS/NCEP/EMC
5830 University Research Ct.
College Park, MD 20740
***@***.***
301-683-3683
—
Reply to this email directly, view it on GitHub
<#2616 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AA4UJHWOJJRC5LTVSEM2MRLZMPVADAVCNFSM6AAAAABJOFZN3KVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTAMBVGI4TENA>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
I found out why setting the show_legend values to True doesn't work. Apparently plotly is expecting 0 or 1 instead of True or False: To show legends, do the following: show_legend:
I'll have to do something in this and the other plotly plots so the behavior is consistent. It isn't very intuitive to set 0/1. |
Beta Was this translation helpful? Give feedback.
-
After this modification, now the legends are presented. See the plot
[image: image.png]
Thanks!
Binbin
…On Mon, Jul 15, 2024 at 12:36 PM bikegeek ***@***.***> wrote:
I found out why setting the show_legend values to True doesn't work.
Apparently plotly is expecting 0 or 1 instead of True or False:
To show legends, do the following:
show_legend:
- 1
- 1
I'll have to do something in this and the other plotly plots so the
behavior is consistent. It isn't very intuitive to set 0/1.
—
Reply to this email directly, view it on GitHub
<#2616 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AQBMPFIA64DJ4SF4K5JWSFTZMP26XAVCNFSM6AAAAABJOFZN3KVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTAMBVGMZTSNQ>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
--
Binbin Zhou
Physical Scientist
Lynker at NOAA/NWS/NCEP/EMC
5830 University Research Ct.
College Park, MD 20740
***@***.***
301-683-3683
|
Beta Was this translation helpful? Give feedback.
-
Sure, thanks!
Binbin
…On Mon, Jul 15, 2024 at 2:19 PM bikegeek ***@***.***> wrote:
I created a Github issue:
dtcenter/METplotpy#455 <dtcenter/METplotpy#455>
Please add anything else in the comments.
—
Reply to this email directly, view it on GitHub
<#2616 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AQBMPFNQIZZJA4ALEOXVS6LZMQHDZAVCNFSM6AAAAABJOFZN3KVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTAMBVGQZDQNA>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
--
Binbin Zhou
Physical Scientist
Lynker at NOAA/NWS/NCEP/EMC
5830 University Research Ct.
College Park, MD 20740
***@***.***
301-683-3683
|
Beta Was this translation helpful? Give feedback.
-
I tested the generation of reliability and ranked histogram but got some issues in the plots:
For the reliability diagram:
(1) if draw plot for just one model, the diagram is OK. But for 2 models, the curve for the model 1 is fine, but the
curve for model 2 has error: the last point (at 90%) and first point (10%) are directly linked.
(2) The location of last point (at 90%) for the model 2 is miss-located. In this case, a triangle polygone.
(3) The legend is shown even though it is sent to "shown"
(4) The colors of the curves for mode1 and model2 are not able to be set (maybe I did not set the colors correctly)
(5) The sharpness plot inside the reliability diagram only has one model. I can not find the option for adding or not adding
the sharpness plot.
For ranked histogram,
(1) one model is fine, but can not put 2 models in one plot (for comparison).
I'll send a completed sample package for testing.
Thanks!
Binbin
EMC/NCEP
Beta Was this translation helpful? Give feedback.
All reactions