Add a devision line in scatter plot of perceptron

I have made a single layer perceptron which will divide a set of points based on the class which is given in the last column of my dataset. I want to visualize the obtained weight vector by a line in the scatter plot of the datapoints. Can anyone help me how I can put the weight vector as a line in the scatter plot? Here's the code for the training dataset

%Data
%Training Data
DataTR=readtable('D:\Universiteit\Hans\Erasmus\Machine Learning\KOM6110_HansHuybrechts_000E19503601_Assignment1\class2_tr.txt');
i_tr=height(DataTR);
X0=ones(i_tr,1);
X1_TR=DataTR.Var1;
X2_TR=DataTR.Var2;
X_TR=[X0 X1_TR X2_TR];
t_TR=DataTR.Var3;
%Training Code
W=rand(3,1);
y_tr=zeros(i_tr,1);
o_TR=zeros(i_tr,1);
dW=ones(3,1);
iterations=1;

%Learning with gradient descent
theta=0.1;
while mean(abs(dW))>0.001
    y_tr=W(1)+ X1_TR.*W(2)+ X2_TR.*W(3);
    o_TR=hardlim(y_tr);
    error(iterations)=1/2*sum((t_TR-o_TR).^2);
    dW=theta*(X_TR.'*(t_TR-o_TR));
    W=W+dW;
    iterations=iterations+1;
end

%Learning with stochastic gradient descent
j=1;
theta=0.005;
while j<=i_tr
    y_tr(j)=W(1)+X1_TR(j)*W(2)+X2_TR(j)*W(3);
    o_TR(j)=hardlim(y_tr(j));
    dW=theta*(X_TR(j,:).'*(t_TR(j)-o_TR(j)));
    W=W+dW;
    j=j+1;
end
gscatter(X1_TR,X2_TR,t_TR,'br','xo',8,'on')

[Ameer Hamza]

Can you attach you dataset. It will help in suggesting a solution.

[Hans Huybrechts]

here's the training data

[Eduardo Morales]

quetal encontraste alguna solucion ?

figure

gscatter(X1_TR,X2_TR,t_TR,'br','xo',8,'on')

hold on

line(x,y)

le agrege esto pero nose si esta bien.

Answer 1

[Jayanti]

Hi Hans,

Once your model is trained, the weight vector W = [w1; w2; w3] defines a linear decision boundary of the form:

On rearranging this to solve for "y" as a function of "x":

We can plot this on top of your scatter plot to visualize the decision boundary.

You can refer to the below code for more details. It creates 100 evenly spaced "x" values and corresponding "y" values using the decision boundary.

% Plotting Data
figure
gscatter(X1_TR, X2_TR, t_TR, 'br', 'xo', 8, 'on')
hold on

% Plotting Decision Boundary Line
x_vals = linspace(min(X1_TR), max(X1_TR), 100);
y_vals = -(W(2)/W(3)) * x_vals - (W(1)/W(3));
plot(x_vals, y_vals, 'k-', 'LineWidth', 2)
hold off

I am also attaching official MathWorks documentation on "plot" for your reference:

https://www.mathworks.com/help/matlab/ref/plot.html