Dear Researchers
I have designed a system for classifying Parkinson patients tremor by UPDRS, but unfortunately I am not sure about the way I am calculating specificity and sensitivity ;-)
I would appreciate if anyone here can help, thanks if your help contain an example
Hi Omid,
Sensitivity of each class can be calculated from its TP/(TP+FN) and
specificity of each class can be calculated from its TN/(TN+FP)
For more information about concept and equations
http://en.wikipedia.org/wiki/Sensitivity_and_specificity
For multi-class classifiacation, you may use one against all approach.
Suppose there are three classes: C1, C2, and C3
"TP of C1" is all C1 instances that are classified as C1.
"TN of C1" is all non-C1 instances that are not classified as C1.
"FP of C1" is all non-C1 instances that are classified as C1.
"FN of C1" is all C1 instances that are not classified as C1.
To find these four terms of C2 or C3 you can replace C1 with C2 or C3.
.
way much simpler (and rougher ...) than the (very interesting but quite theoretical !) reference above :
http://rduin.nl/papers/prasa_06_vuc.pdf
.
Dear Omid,
You can use the attached file to calculate classification performance statistical measures.
Every week I get notifications like so many people viewed this question that I asked 3 years ago. I thought if I share this link with you, it would be helpful. In this link, very good information are provided in series of short videos regarding EVALUATION. I hope it helps you dear researchers: https://www.youtube.com/watch?v=M4tfAal7ElQ&list=PLBv09BD7ez_50pj5kcKFYee7QPcg3ImCV
In addition to the series videos have been taken in the link above by Omid,this link may reduce your difficulties about this problem https://www.youtube.com/watch?v=FAr2GmWNbT0
Dear Researchers,
Can I take the average of each class' sensitivity as the overall sensitivity of the classifier?
Let me ask you a more general question, please.
I'm trying to evaluate a multi-class classifier using the ROC Curve.
Given the input data, the ranker returns a number for each of my five classes. The classifier determines the class of such data by choosing the class with the highest score number. So, it does not consider that a data belongs to a class by just checking if the score is above a cutoff threshold as it occurs in a binary classifier problem, for example.
My question is: In order to create a ROC Curve, how can a choose a threshold to vary during the experiments in such a multi-class ranking classifier problem? Is it possible?
Thank you.
Article Understanding and using sensitivity, specificity and predictive values
Thank you Bazhdar Nouredine. The video explained the context clear enough.
Dear Omid Bazgir
These articles will clarify your question clearly
Article Novel Feature Selection and Voting Classifier Algorithms for...
Article MbGWO-SFS: Modified Binary Grey Wolf Optimizer Based on Stoc...
Dear Omid Bazgir
These articles will clarify your question clearly
Article Novel Feature Selection and Voting Classifier Algorithms for...
Article MbGWO-SFS: Modified Binary Grey Wolf Optimizer Based on Stoc...
Refer to this article, please:
A systematic analysis of performance measures for classification tasks.
Sokolova, Marina
Lapalme, Guy
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts