how to compare the class of input and output,and display the misclassification,how much percentage it is classified properly

3 views (last 30 days)
x1 x2 class
a= -1.7986 -1.6730 1.0000
-1.0791 -0.5937 1.0000
-0.5995 0.7556 1.0000
1.0791 -1.4032 1.0000
0.1199 0.2159 1.0000
0.3597 0.4857 -1.0000
-0.3597 1.5651 -1.0000
0.5995 0.4857 -1.0000
0.1199 -0.3238 -1.0000
1.5588 0.4857 -1.0000
result=x1 x2 wx-gamma class
-1.7986 -1.6730 0.8068 1.0000
-1.0791 -0.5937 0.3781 1.0000
-0.5995 0.7556 -0.0706 -1.0000
1.0791 -1.4032 0.1382 1.0000
0.1199 0.2159 -0.0808 -1.0000
0.3597 0.4857 -0.2004 -1.0000
-0.3597 1.5651 -0.3298 -1.0000
0.5995 0.4857 -0.2503 -1.0000
0.1199 -0.3238 0.0588 1.0000
1.5588 0.4857 -0.4500 -1.0000

Answers (1)

Ahmed
Ahmed on 30 Sep 2014
To just get the accuracy it is only required to count the number of matches and divide by the total number of observations:
acc = sum(a.class == result.class)/size(a.class,1),
However, you should consider having a look at the confusion matrix as well:
cfMat = confusionmat(a.class,result.class),
acc = sum(diag(cfMat))/sum(cfMat(:)),
Then print the result nicely:
fprintf('Accuracy: %.1f%%\n',100*acc);
In addition, investigating some sort of performance curve is also helpful:
[FPR,TPR,~,AUC] = perfcurve(a.class, result.wx_gamma,1);
plot(FPR,TPR);
axis('equal');
axis([0 1 0 1]);
hold on; grid on;
line([0 1],[0 1]);
hold off;
xlabel('FPR'); ylabel('TPR');

Categories

Find more on Statistics and Machine Learning Toolbox in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!