clc
clear
warning('off','all')
for n=1:5
switch n
    case 1
       fisdosyaa='gaussmf1.fis';filename1 = 'PredTrainV1.xlsx';filename2 = 'PredTestV1.xlsx';
       case 2
        fisdosyaa='gaussmf2.fis';filename1 = 'PredTrainV2.xlsx';filename2 = 'PredTestV2.xlsx';
          case 3
        fisdosyaa='gaussmf3.fis';filename1 = 'PredTrainV3.xlsx';filename2 = 'PredTestV3.xlsx';
        case 4
        fisdosyaa='gaussmf4.fis';filename1 = 'PredTrainV4.xlsx';filename2 = 'PredTestV4.xlsx';
    otherwise 
       fisdosyaa='gaussmf5.fis';filename1 = 'PredTrainV5.xlsx';filename2 = 'PredTestV5.xlsx';
end

datatrn = xlsread(filename1); % Read training values from the Excel file
datatest = xlsread(filename2); % Read test values from the Excel file

% assign the real data to the variable from training dataset 
real_column = datatrn(:, 1);datatrn(:,1)=[];
% assign the real data to the variable from testing dataset 
real_columntest = datatest(:, 1);datatest(:,1)=[];
% Calculate MAE for each column in training dataset
mae_values = zeros(1, size(datatrn, 2)); % Excluding the last column
for i = 1:(size(datatrn, 2))
    mae_values(i) = calculateMAE(datatrn(:, i), real_column);
end

% Find the column with the best MAE and assign it as input 1 of ANFIS
[best_mae, best_column] = min(mae_values);
input1=datatest(:, best_column);% best predictor
% Select data with the smallest MA in the other four columns in testin dataset, and assign it
% as input 2 of ANFIS
other_columns  = setdiff(1:size(datatest, 2), best_column); % Exclude the best column
selected_data  = zeros(1, length(other_columns));

    for j = 1:length(datatest)
        for i = 1:length(other_columns)
            selected_data(i) = abs(datatest(j, other_columns(i)) - input1(j,1));
        end
        [minMA, minMAIndex] = min(selected_data);
        input2(j)=datatest(j,minMAIndex);
    end
Finputs=[input1 input2' real_columntest];
    % Combine the best two entries and get the last output.
anfis=readfis(fisdosyaa);%**************************************************************/  
anfis_output = evalfis(anfis,Finputs(:,1:2));%*******************************************/
% Calculate fold performances
mae = mean(abs(anfis_output - real_columntest));% Calculate MAE
rmse = sqrt(mean((anfis_output - real_columntest).^2));% Calculate RMSE
mse = mean((anfis_output - real_columntest).^2);% Calculate MSE
mape = mean(abs((real_columntest - anfis_output) ./ real_columntest));% Calculate MAPE
mae1(n)=mae;rmse1(n)=rmse;mse1(n)=mse;mape1(n)=mape;
fprintf('Fold %1.0f',n); fprintf(' Performance Results\n');
fprintf('MAE....: %.8f', mae);fprintf('   RMSE...: %.8f', rmse);
fprintf('   MSE....: %.8f', mse);fprintf('   MAPE...: %.8f\n', mape);
%/*****************************%/**************************************************************/
end