nnetar giving different output each time

Question

I am using nnetar package for time series forecasting in R.But each time when I run the model, I am getting different output. i guess this is related to different random weight initialization at each time, but I don't know how to set it as constant.

ts=nnetar(Time_Series,P=3,lambda = 0,decay=0.001,maxit=150)
fs = forecast(ts, h=12)

Any pointers would be appreciated.

Answer 1

Just set the random seed before you call nnetar and you will get the same random initialization each time.

set.seed(12345)
ts=nnetar(Time_Series,P=3,lambda = 0,decay=0.001,maxit=150)
fs = forecast(ts, h=12)