you can check this link for calculation the metrics of cohen_kappa_score for CNN model

https://scikit-learn.org/stable/modules/generated/sklearn.metrics.cohen_kappa_score.html

Dear Hussain,

please go through this simple ppt. You will get good idea.

Dear Muhammad,

Kappa Coefficient, a.k.a. Cohen's kappa can be easily calculated using a formula and the number of true positive, false positive, false negative and true positive cases from the confusion matrix (contingency table). You can follow the example of this wikipedia-page:

https://en.wikipedia.org/wiki/Cohen%27s_kappa

If you use R statistical software, the most straightforward way calculating kappa by function cohen.kappa() of package 'psych', function Kappa() of package 'vcd' or function Kappa.test() of package 'fmsb'.

Note that the function kappa() of package 'base' is not for Cohen's kappa but for calculation of the Condition Number (multicollinearity measure).

HTH,

Ákos

Simple!!!!! If you are remote sensing specialist ,you can use software like ENVI and calculat it automatically via accuracy assessment tool

There is a very nice book on computation of kappa and Tau by Congalton and Green. This book also has statistical z-test analysis for error matrices which are significant for understanding the comparison between multiple classifiers.

Dear

they are an accuracy parameters, and to get them you need to collect ground truth data and compared to your data

Article Death to Kappa: Birth of quantity disagreement and allocatio...

The kappa statistic is used to control only those instances that may have been correctly classified by chance. This can be calculated using both the observed (total) accuracy and the random accuracy. Kappa can be calculated as:

Kappa = (total accuracy – random accuracy) / (1- random accuracy).

Article Development of a Data Mining Based Model for Classification ...

Sourabh Shastri Arun Singh Bhadwal Paramjit Kour

n essence, the kappa statistic is a measure of how closely the instances classified by the machine learning classifier matched the data labeled as ground truth, controlling for the accuracy of a random classifier as measured by the expected accuracy.

thank you all

you can check this useful link, it explains the different methods to assess the accuracy of the classification of remotely sensed data: http://www.50northspatial.org/classification-accuracy-assessment-confusion-matrix-method/