Hello, I would like some help with matlap please. I was given this script and I would like a matlab MLP which follows the same inputs and outputs of this question so that I can see an example of how MLP's work in matlab!

It takes in a big excel sheet with 3 classes and then it takes another for testing. It then outputs 3 outputs. I just need a script that will run better than this base one!

% ########################## Assignment 5 base script

% needs to be in the same directory as the data files

numClass = 3;

data = uiimport('train.csv');

size(data.train)

input = data.train;

size(input)

input(:,43 ) = []; % Remove class labels

size(input)

d = data.train(:,43); % Stroe class labels in d

x = input';

d = d';

% Get the class distribution in training set (display purposes only)

nc = zeros(1,numClass)

for i = 1:numClass

nc(1,i) = sum(d(:) == i);

disp(sprintf('Class %d : %d', i, nc(1,i)))

end

% Create binary matrix of class labels ************************************

% There are now three outputs, i.e. one output per class

d = dummyvar(d);

d = d';

% Normalize inputs ********************************************************

[xn, xs] = mapminmax(x); % normalize inputs

% % Basic linear network, should give ADR of ~0.5 using train.csv, and ~0.59 using trainOversampled.csv ***

maxlr = 0.40*maxlinlr(xn); % initialize learning rate and network weights

net1 = newlin(minmax(xn), numClass, [0], maxlr); % one output for each class

net1.trainParam.epochs = 6000;

[net1,tr] = train(net1,xn,d);

% Establish output of neuron and plot the confusion matrix ****************

y = sim(net1,xn);

plotconfusion(d,y);

% Import the test data (no label information) *****************************

data = uiimport('test.csv');

input = data.test;

x = input';

% Reapply the normalization estimated on the training set *****************

xnTest = mapminmax('apply', x, xs);

% Perform forward pass through the network under TEST data ****************

y = sim(net1, xnTest);

csvwrite('B00708156-test.csv',y); % Save network outputs