How do I improve my result of KNN classification using confusion matrix?

Hello everyone.

I'm trying to classify a data set containing two classes using a Knn classifer.

and would like to evaluate the performance using its confusion matrix. But how can I use it with the KNN classifier?

This is my code of KNN classifer

model=ClassificationKNN.fit(X,Y,'NumNeighbors',9);
[~,result1]=predict(model,x);

[Image Analyst]

You forgot to attach X and Y in a .mat file

save('answers.mat', 'X', 'Y');

Have you tried the "Classification Learner" App on the App tab of the tool ribbon?

You tagged it with image processing. What about this is at all related to image processing???

[youb mr]

how i can use confusion_matrix in this situation

Answer 1

[Ridwan Alam]

yhat = predict(model,x);
[C,order] = confusionmat(y,yhat);

Use this help file to understand how to use C and order:

https://www.mathworks.com/help/stats/confusionmat.html

[youb mr]

Error using confusionmat (line 98)

G and GHAT need to have same number of rows

Error in knn (line 189)

C = confusionmat(Y,yhat)

[Ridwan Alam]

Here, I am assuming you have trained the model with “X” and “Y”, and are testing with “x” and “y”. “X” and “x” are different data, if in matrix format, they should have same number of columns but different row sizes.

“yhat” is the prediction of your model for test data “x” (not “X”). Confusionmat compares “yhat” with the ground truth or labels “y” (not “Y”) for the test data “x”.