Forums

Back

Confidence intervals

RSS (Opens New Window)(Opens New Window)
AO
Alessandro Orio, modified 3 Years ago.

Confidence intervals

Youngling Posts: 20 Join Date: 5/13/19 Recent Posts

Hi everyone!

 

I have a question regarding how to calculate correctly the confidence intervals for F, Recruitment and SSB.

If I look at the time series plots coming from the SS_plots functions it looks like the CI for F and SSB are symmetrical while the ones of Recruitment are not. Instead, the CI of the recruitment plot coming from the function SSplotComparisons are different and look symmetrical.

I was wondering what is the difference between the two CI calculated in the different functions? And what is the correct way to either extract them in r4ss or to calculate them?

 

Thanks in advance for the help,

Ale

CW
Chantel Wetzel, modified 3 Years ago.

RE: Confidence intervals

Youngling Posts: 32 Join Date: 10/7/15 Recent Posts
The confidence interval around SB and F is normally distributed and the confidence interval around recruitment is log-normal.  Each are calculated as:

recruitment interval =  exp(log(value) -/+ qnorm(1-(1-CI)/2) * sqrt(log(1 + (sd/value) * (sd/value))))  

SB or F interval =  value - qnorm(1-(1-CI)/2)*sd    

where value is the MLE estimate, CI is the confidence interval, and sd is the standard deviation.  The confidence interval used in the r4ss figures is set at 95%.

The MLE estimated value and the standard deviation for each derived quantity can be found in the Report file under the columns labeled "Value" and "StdDev" respectively.


Chantel Wetzel, PhD

Pronouns: she/her (what does that mean?)

Fishery Resource Analysis and Monitoring Division
Northwest Fisheries Science Center
National Marine Fisheries Service
National Oceanic and Atmospheric Administration
Office: 206.302.1753 


On Tue, Sep 1, 2020 at 7:03 AM Alessandro Orio <VLab.Notifications@noaa.gov> wrote:

Hi everyone!

 

I have a question regarding how to calculate correctly the confidence intervals for F, Recruitment and SSB.

If I look at the time series plots coming from the SS_plots functions it looks like the CI for F and SSB are symmetrical while the ones of Recruitment are not. Instead, the CI of the recruitment plot coming from the function SSplotComparisons are different and look symmetrical.

I was wondering what is the difference between the two CI calculated in the different functions? And what is the correct way to either extract them in r4ss or to calculate them?

 

Thanks in advance for the help,

Ale


--
Alessandro Orio Stock Synthesis Virtual Lab Forum https://vlab.noaa.gov/web/stock-synthesis/public-forums/-/message_boards/view_message/11664630 VLab.Notifications@noaa.gov
AO
Alessandro Orio, modified 3 Years ago.

RE: Confidence intervals

Youngling Posts: 20 Join Date: 5/13/19 Recent Posts

Thank a lot Chantel.

So what about the Recruitment CI in SSpltocomparison function? Is it a bug?

 

Thanks again for the quick reply!

 

Ale

CW
Chantel Wetzel, modified 3 Years ago.

RE: Confidence intervals

Youngling Posts: 32 Join Date: 10/7/15 Recent Posts
Hi Ale

I think I have likely looked at thousands of SSplotcomparison figures but have never noticed that the recruitment interval was not log-normal here.  After digging through the r4ss code, it looks like the SSsummarize function (creates the values passed to the SSplotcomparison function) is calculating the interval around recruitment as normally distributed which is incorrect and inconsistent with the SS_plots function.  We will get this corrected.  Thank you for catching this issue.

Chantel Wetzel, PhD

Pronouns: she/her (what does that mean?)

Fishery Resource Analysis and Monitoring Division
Northwest Fisheries Science Center
National Marine Fisheries Service
National Oceanic and Atmospheric Administration
Office: 206.302.1753 


On Tue, Sep 1, 2020 at 7:23 AM Alessandro Orio <VLab.Notifications@noaa.gov> wrote:

Thank a lot Chantel.

So what about the Recruitment CI in SSpltocomparison function? Is it a bug?

 

Thanks again for the quick reply!

 

Ale


--
Alessandro Orio Stock Synthesis Virtual Lab Forum https://vlab.noaa.gov/web/stock-synthesis/public-forums/-/message_boards/view_message/11665056 VLab.Notifications@noaa.gov
AO
Alessandro Orio, modified 3 Years ago.

RE: Confidence intervals

Youngling Posts: 20 Join Date: 5/13/19 Recent Posts

Thanks to you for the all the effort you put in answering our questions!

Ale

 

From: Chantel Wetzel <VLab.Notifications@noaa.gov>
Sent: 01 September 2020 17:06
To: VLab.Notifications@noaa.gov
Subject: [Stock Synthesis - SS usage questions] Confidence intervals

 

Hi Ale

 

I think I have likely looked at thousands of SSplotcomparison figures but have never noticed that the recruitment interval was not log-normal here.  After digging through the r4ss code, it looks like the SSsummarize function (creates the values passed to the SSplotcomparison function) is calculating the interval around recruitment as normally distributed which is incorrect and inconsistent with the SS_plots function.  We will get this corrected.  Thank you for catching this issue.

 

Chantel Wetzel, PhD

 

Pronouns: she/her (what does that mean?)

 

Fishery Resource Analysis and Monitoring Division

Northwest Fisheries Science Center

National Marine Fisheries Service

National Oceanic and Atmospheric Administration

Office: 206.302.1753 

 

 

On Tue, Sep 1, 2020 at 7:23 AM Alessandro Orio <VLab.Notifications@noaa.gov> wrote:

Thank a lot Chantel.

So what about the Recruitment CI in SSpltocomparison function? Is it a bug?

 

Thanks again for the quick reply!

 

Ale


--
Alessandro Orio Stock Synthesis Virtual Lab Forum https://vlab.noaa.gov/web/stock-synthesis/public-forums/-/message_boards/view_message/11665056 VLab.Notifications@noaa.gov


--
Chantel Wetzel Stock Synthesis Virtual Lab Forum http://vlab.noaa.gov/web/stock-synthesis/public-forums/-/message_boards/view_message/11665208 VLab.Notifications@noaa.gov

---
När du skickar e-post till SLU så innebär detta att SLU behandlar dina personuppgifter. För att läsa mer om hur detta går till, klicka här
E-mailing SLU will result in SLU processing your personal data. For more information on how this is done, click here

U
Anonymous, modified 3 Years ago.

RE: Confidence intervals

Hi Ale,

The issue is now resolved in the "development" branch of r4ss available via 

remotes::install_github("r4ss/r4ss", ref = "development")

There was a github issue created for this a few years ago at https://github.com/r4ss/r4ss/issues/181 but it never got acted on until now, so thanks for speaking up.

-Ian

AO
Alessandro Orio, modified 3 Years ago.

RE: Confidence intervals

Youngling Posts: 20 Join Date: 5/13/19 Recent Posts

Thank a lot Ian!

Best,

Ale

 

From: Anonymous <VLab.Notifications@noaa.gov>
Sent: 01 September 2020 19:18
To: VLab.Notifications@noaa.gov
Subject: [Stock Synthesis - SS usage questions] Confidence intervals

 

Hi Ale,

The issue is now resolved in the "development" branch of r4ss available via 

remotes::install_github("r4ss/r4ss", ref = "development")

There was a github issue created for this a few years ago at https://github.com/r4ss/r4ss/issues/181 but it never got acted on until now, so thanks for speaking up.

-Ian


--
Anonymous Stock Synthesis Virtual Lab Forum https://vlab.noaa.gov/web/stock-synthesis/public-forums/-/message_boards/view_message/11667307 VLab.Notifications@noaa.gov

---
När du skickar e-post till SLU så innebär detta att SLU behandlar dina personuppgifter. För att läsa mer om hur detta går till, klicka här
E-mailing SLU will result in SLU processing your personal data. For more information on how this is done, click here