My Code for Input-Output Mapping by Neural Network toolbox (nftool) in MATLAB is correct?

I have written a code for Mapping using nftool and its as follow.

%-----------------------------------------------------------------

clc

clear all

load input.mat % Size is [1*200]

load output.mat % Size is [1*200]

load newinput.mat % Size is [1*5]

x=input;

t=output;

x1=newinput;

trainFcn = 'trainlm'; % Training algorithm

hiddenLayerSize = 5; % No. of hidden neurons

net = fitnet(hiddenLayerSize,trainFcn); % Network is developed

[trainInd,valInd,testInd]=divideind(200,1:150,151:180,181:200) % Dividing samples

net.divideParam.trainInd= 1:150

net.divideParam.valInd= 151:180

net.divideParam.testInd= 181:200

[net,tr] = train(net,x,t);

y = net(x);

e = gsubtract(t,y);

performance = perform(net,t,y);

view(net);

y1=net(x1) % Output for newinput x1

%------------------------------------------------------------------------------------

I have an input data of [1*200] in size and corresponding target(output) data of [1*200]. I am using " divideind " to divide my samples for developing the network. to test the mapping I have a test data named " newinput ". My samples, newinput and number of neurons are fixed. When I re-run the code I get different answer of " y1 " for same input " x1 ". 

So my question is " Why I am getting different answers for " y1 " when my all inputs are constant? ". Is anything wrong with my code?