k nearest neighbour classifier Matlab

Pattern Recognition

K-Nearest Neighbor Classification

K nearest neighbors is a simple algorithm that stores all available cases and classifies new cases based on a similarity measure (e.g., distance functions).

 Example 

 CODES 

clear;

hold on;

axis equal;

axis([0 1 0 1]);

for i=1:10;

x(i,:)=ginput(1);

text(x(i,1), x(i,2), '1');

plot(x(i,1), x(i,2), '.');

end;

for i=11:20;

x(i,:)=ginput(1);

text(x(i,1), x(i,2), '2');

plot(x(i,1), x(i,2), '.');

end;

for j=1:4;

display('please input new sample with mouse');

x1(j,:)=ginput(1)

for i=1:20; dx(i)=norm(x1(j,:)- x(i,:));

end;

[xd,id] = sort(dx);

kd=floor(id(1:3)/10-1e-3)+1 ;

counter1=0;

counter2=0

for n=1:3;

if kd(n)==1 

counter1=counter1+1;

else

counter2=counter2 +1;

end

end;

if counter1>=2

class=1

else

class=2

end

switch class

case 1

M = 1

case 2

M = 2

otherwise

error('This is impossible')

end

switch j

case 1

s = 'A';

case 2

s = 'B';

case 3

s = 'C';

case 4

s = 'D';

otherwise

error('This is impossible')

end

text(x1(j,1), x1(j,2), s)

plot([x1(j,1) x(id(1),1)],[x1(j,2) x(id(1),2)],'-');

plot([x1(j,1) x(id(2),1)],[x1(j,2) x(id(2),2)],'-');

plot([x1(j,1) x(id(3),1)],[x1(j,2) x(id(3),2)],'-');

end;