Re: [R] Sensitivity and Specificity

[Prof. Dr. Matthias Kohl]

MKclass::perfMeasures(predict_testing, truth = testing$case, namePos = 1)

should also work and computes 80 performance measures.

Best Matthias

Am 25.10.22 um 06:42 schrieb Jin Li:

Hi Greg,

This can be done by:
spm::pred.acc(testing$case,  predict_testing)

It will return both sensitivity and specificity, along with a few other
commonly used measures.

Hope this helps,
Jin

On Tue, Oct 25, 2022 at 6:01 AM Rui Barradas <ruipbarra...@sapo.pt> wrote:

Às 16:50 de 24/10/2022, greg holly escreveu:

Hi Michael,

I appreciate your writing. Here are what I have after;

predict_testing <- ifelse(predict > 0.5,1,0)

head(predict)

           1          2          3          5          7          8
0.29006984 0.28370507 0.10761993 0.02204224 0.12873872 0.08127920

# Sensitivity and Specificity

sensitivity<-(predict_testing[2,2]/(predict_testing[2,2]+predict_testing[2,1]))*100

Error in predict_testing[2, 2] : incorrect number of dimensions

sensitivity

function (data, ...)
{
      UseMethod("sensitivity")
}
<bytecode: 0x000002082a2f01d8>
<environment: namespace:caret>

specificity<-(predict_testing[1,1]/(predict_testing[1,1]+predict_testing[1,2]))*100

Error in predict_testing[1, 1] : incorrect number of dimensions

specificity

function (data, ...)
{
      UseMethod("specificity")
}
<bytecode: 0x000002082a2fa600>
<environment: namespace:caret>

On Mon, Oct 24, 2022 at 10:45 AM Michael Dewey <li...@dewey.myzen.co.uk>
wrote:

Rather hard to know without seeing what output you expected and what
error message you got if any but did you mean to summarise your variable
predict before doing anything with it?

Michael

On 24/10/2022 16:17, greg holly wrote:

Hi all R-Help ,

After partitioning my data to testing and training (please see below),

I

need to estimate the Sensitivity and Specificity. I failed. It would be
appropriate to get your help.

Best regards,
Greg


inTrain <- createDataPartition(y=data$case,
                                  p=0.7,
                                  list=FALSE)
training <- data[ inTrain,]
testing  <- data[-inTrain,]

attach(training)
#model training and prediction
data_training <- glm(case ~ age+BMI+Calcium+Albumin+meno_1, data =
training, family = binomial(link="logit"))

predict <- predict(data_training, data_predict = testing, type =

"response")

predict_testing <- ifelse(predict > 0.5,1,0)

# Sensitivity and Specificity

  
sensitivity<-(predict_testing[2,2]/(predict_testing[2,2]+predict_testing[2,1]))*100

    sensitivity

  
specificity<-(predict_testing[1,1]/(predict_testing[1,1]+predict_testing[1,2]))*100

    specificity

        [[alternative HTML version deleted]]

______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide

http://www.R-project.org/posting-guide.html

and provide commented, minimal, self-contained, reproducible code.

--
Michael
http://www.dewey.myzen.co.uk/home.html

       [[alternative HTML version deleted]]

______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide

http://www.R-project.org/posting-guide.html

and provide commented, minimal, self-contained, reproducible code.

Hello,

Instead of computing by hand, why not use package caret?


tbl <- table(predict_testing, testing$case)
caret::sensitivity(tbl)
caret::specificity(tbl)


Hope this helps,

Rui Barradas

______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

--
Prof. Dr. Matthias Kohl
www.stamats.de

______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.