RBF- validation and training set

clc
clear all
close all
format long
out = csvread('Data_1000points_sampled_every5.csv');%importing the data file
u= [out(:,1) out(:,2)]'; % matrix containing u1 and u2
xo = out(:,3)'; %matrix containing x1
time=out(:,4)'; % time stamps
net=0;
[row_u,col_u]=size(u); %getting size of dtatset
[row_xo,col_xo]=size(xo);

training=[];
training=[u(1,1:col_u-100);u(2,1:col_u-100)];
[row_training,col_training]=size(training);
training_time=[time(1,1:col_u-100)];

validation=[];
validation=[u(1,col_u-99:col_u);u(2,col_u-99:col_u)];
[row_validation,col_validation]=size(validation);
validation_time=[time(1,col_u-99:col_u)];

column_num=1;%keeps track of column index
weighted_in =[];%matrix to store 5 u1 and u2 values from orignal dataset
weighted_out=[];%stores corresponding value of x1

simulate=[];% stores the value of u1 and u2 to be used in sim function
column_num1=1;% keeping track of column index
sim_ans=[];% stores the simulated value of x1 generated by sim function
error=[];
error_valid=[];
% how many elements we want to look back on
index=1;
Mean_error=[];
sim_valid_ans=[];
%  for a=1:10
num_of_elements=5;
column_num1 = 1;
error=[];

for i=1:col_training-num_of_elements % outer loop
    column_num=1;
    for j=i:i+num_of_elements-1% inner loop
        %in this loop we generate our smaller datasets and feed them into
        %the rbe
        
        weighted_in(1,column_num)=u(1,j);
        weighted_in(2,column_num)=u(2,j);
        
        weighted_out(1,column_num)=xo(1,j);
        
        column_num=column_num+1;
    end
    % feeding it into the rbe
    net=newrbe(weighted_in,weighted_out);
    
    % now we simulate the value of x1 in the timestep that is "next" and hasnt
    % been input into the net yet
    simulate(1,1)=training(1,i+num_of_elements);% getting the values of u1 and u2 at the "next" timestep
    simulate(2,1)=training(2,i+num_of_elements);
    sim_ans(1,column_num1) = sim(net,simulate);% storing the simulated value
    sim_ans(2,column_num1)=training_time(1,i);% storing the corresponding timestep
    
    if xo(1,i)==0
        error(1,column_num1)=0;
        column_num1 = column_num1+1;
    else
        error(1,column_num1)=abs(sim_ans(1,column_num1)-xo(1,i))/xo(1,i);
        column_num1 = column_num1+1;
    end
    

end
%simulating
column_num1=1;
column_num=1;
for x=1:col_validation-num_of_elements
%     column_num=1;
%    for y=x:x+num_of_elements-1
%     p(1,column_num)= validation(1,y)
%     p(2,column_num)= validation(2,y)
%     column_num=column_num+1;
%    end
   p=[validation(1,x);validation(2,x)];
    sim_valid_ans(1,column_num1)=sim(net,p);
   
    if xo(1,x+col_training)==0
        error_valid(1,column_num1)=0;
        column_num1 = column_num1+1;
    else
        error_valid(1,column_num1)=abs(sim_valid_ans(1,column_num1)-xo(1,x+col_training))/xo(1,x+col_training);
        column_num1 = column_num1+1;
    end
    sim_valid_ans(2,column_num1)=validation_time(1,x);% storing the corresponding timestep
end

% plotting orignal data and simulated data

% Mean_error(1,index)= mean(error,'all')*100;
% Mean_error(2,index)= index;
% index=index+1;
% legend('
%  end

% figure
% plot( Mean_error(2,:),Mean_error(1,:))
%      figure

 plot( time,xo(1,:),'g',sim_ans(2,:),sim_ans(1,:),'b')
 legend("Actual data(training and validation combined)","simulation for training only")
 title('Actual data and simulation training set ')
 xlabel('Time')

 figure
 plot(sim_ans(2,:), error);
title('error graph training')
xlabel('Time')

figure 
plot(validation_time,xo(1,col_training+1:col_u),'g',sim_valid_ans(2,:), sim_valid_ans(1,:));
legend('actual data', ' simulated');
title(' actual and simulated for validation set');