Artificial neural network (ANN)

my study is the performance comparison between four networks training functions methods in Artificial neural network (ANN), trainlm, traingda trainbfg and trainrp in matlab language I use two test Complex modulus and phase angle I use this cod this cod ------------------------------------------------------------- net=feedforwardnet(10,'trainlm'); net=train(net,x,t); p=sim(net,i); ------------------------------------------------------------------ I show here where I put 'trainlm or traingda trainbfg and trainrp net=feedforwardnet(10,'trainlm or traingda trainbfg and trainrp '); net=train(net,x,t); p=sim(net,i); before the running the program must it get three variables x,t,and i the data it divided by two parts testing and training it will be like that 70% of input and output these are training the 70% of input it is called (x) the 70% of output it called (t) . 30% of output is testing data this one called (i) . The 30% of output it is the actual data that it will be compare between it and the predicted data that it we get it after program running and after using the below functions to get the errors RMSE, MAE MSE and R^2 as show (MAE=1/n ∑_(i=1)^n▒|x-y| ) (MSE=1/n ∑_(i=1)^n▒〖(x-y)〗^2 ) X is actual data Y is predate data After I run get this I fund good regression but the problem I faced is the MAE MSE RMSE are not good there is very big different after I compere the actual and the predicted data I use data it content three types of asphalt those come from three types of petrol middle east, Russian and Venezuelan and 3 angles (3,5and 7) degree I arraignment the data All data for phase angle complex modules 1008 sample for modify test every test All data for phase angle complex modules 322 sample for control test every test and I sort it from smallest to the lager number the first three Columns Are the input( the first one is freq second one temperature third one the angel of twist fourth one the types of petrol ) I defined the types by number 1,2 and 3 Number 1 Middle East Number 2 Russian Number 3 Venezuelan last one is the output(after sort it in Microsoft Excel) I use Microsoft Excel to find the errors by Excel equations this a a sample from the data 1.5 65 3 1 2760 0.1 45 3 2 2780 1 65 5 1 2800 0.1 55 7 2 2885 0.015 45 7 1 2900 0.5 55 3 3 2930 0.2 55 5 3 2935 0.1 55 7 3 2950 0.1 45 3 3 3225 0.015 35 3 2 3350 i want some one help me to less err between actual and the predicted data by Writing a cod or give me any way in ANN to get less erros in matlab language