Problem in forecasting using GRNN

I will predict some values with GRNN, but I have problem with output values. All output values of testing have same value with last value of output in training.

x =[1 2 3 4 5 6 7 8 9 10] ;

y =[1 4 6 8 10 12 14 16 18 20] ;

xx = 30:35;

[x, map] = mapstd(x);

xx = mapstd('apply', xx, map);

y = mapstd('apply', y, map);

s = 0.3;

grnn = newgrnn(x, y, s);

test_P = sim(grnn, xx);

test_P = mapstd('reverse',test_P,map);

display(test_P)

>>tes

test_P =

20 20 20 20 20 20

Answer 1

[Ulin Nuhaq]

does anyone know the answer above?

Answer 2

[ahmed ayad]

the same problem is with me>>>> i think that grnn is not appropraite for prediction>>>

Answer 3

[Shubham Rawat]

Hi Ulin,

While you are creating network

grnn = newgrnn(x,y,s); 

You have used new x and y after changed using “mapstd”, instead of that you may use original values of x and y. That will lead you to desired results.