function InferenceEngineandDefuzzification(combination,trainingdata,validationdata,testdata,knowledgeRuleBase,indicesinputs,indicesoutputs)

fis = newfis(put the fis file name);

define the defuzzification method;

for i = 1: size(indicesinputs,2)

put input variable name;

assign ranges for the input variables;

for j = 1: inputs(indicesinputs(combination,i)

put name to the membership function for the input variable

put type to the membership function for the input variable

if type == ‘trapmf’

put parameters = knowledgeDataBase(j,:,indicesinputs(combination,i));

else

put parameters = knowledgeDataBase(j,[1,2,4],indicesinputs(combination,i));

end

end

end

for i = 1: size(indicesoutputs,2)

put output variable name;

assign ranges for the output variables;

for j = 1: inputs(indicesoutputs(combination,i)

put name to the membership function for the output variable

put type to the membership function for the output variable

if type == ‘trapmf’

put parameters = knowledgeDataBase(j,:,indicesoutputs(combination,i));

else

put parameters = knowledgeDataBase(j,[1,2,4],indicesoutputs(combination,i));

end

end

end

for m = 1: size(knowledgeRuleBase,1)

put antecedents;

put consequents;

put weights;

put connections;

end

namevars = concatenate(‘Vars_’,indicesinputs(combination,:));

[performances(:,:,combination)] = CrispOutputs(fis,trainingdata,validationdata,testingdata,indicesinputs,indicesoutputs,combination);

namefis = concatenate(performances(:,:,combination),’#_vars’,size(indicesinputs,2),’_’,namevars);

save the fis file: ‘*.fis’ with the namefis variable;

save the variables: ‘*.mat’ with the namefis variable;

end