How to prepare the data for Classification learner Toolbox?

Dears,

I am doing parking lot detection by using Matlab tool called classification learner. I prepared the dataset as tables and imported it to the classifier but unfortunately, I got training model with 75% accuracy.

I do not know how to improve the classifier accuracy or and how to get rid of variables overlap in scatter plot?

is the problem with my data or in the way I prepared it ?

here is the code for Image Batch processor tool:

function occupied = myimfcn(im)
%Input image
im=rgb2gray(im);
J=imresize(im, [30 80]);
B = reshape(J,1,[]);
occupied.occupied =B;

preparing data code. allresults(1,i).occupied The struct array I got from Image Batch processor tool

for i=1:1000
C(i,:)=allresults(1,i).occupied;
end
A= repmat('empty',500,1);
B = repmat('occupied',500,1);
[sA1, sA2] = size(A);
[sB1, sB2] = size(B);
L(sA1+1:sA1+sB1, 1:sB2) = B;
L(1:sA1, 1:sA2)         = A;
D = array2table(C);
D= addvars(D,L,'After','C2400');

Answer 1

[Santhosh A V]

Hi Abdulaziz Alghamdi

Data preparation for training seems good. You can refer to “Compare and Improve Classification Models” section in below link to improve the accuracy of the classifier. https://in.mathworks.com/help/stats/train-classification-models-in-classification-learner-app.html#bu3xf0x-1

Hope this helps.

Thanks

Santhosh

[Abdulaziz Alghamdi]

Thanks Santhosh I really appreciate your response.