Hi,
I have a time-series with 104 values. Measurement taken by an instrument, every 6 months
The auto.arima() uses a seasonal model to forecast future values. If I decompose in trend, seasonal, and random, I can see that the seasonal component is very small [between -0.04 and 0.04], similar to random values.. and I know that the precision of the instrument used is +-0.1.
My question is.. can we assume that there is no seasonal component as what I see is smaller than random noise? and the just use a non-seasonal ARIMA?
Is it correct to consider the seasonal component, or is it just noise?
I attach a picture and the values... thanks!
0.00 -0.18 -0.16 -0.46 -0.21 -0.44 -0.19 -0.29 -0.16 -0.34 -0.30 -0.45 -0.04 -0.44 -0.34 -0.38 -0.35 -0.39 -0.24 -0.37 -0.36 -0.58 -0.43 -0.46 -0.48 -0.54 -0.37 -0.44 -0.40 -0.45 -0.42 -0.35 -0.30 -0.46 -0.46 -0.48 -0.26 -0.46 -0.35 -0.47 -0.33 -0.50 -0.50 -0.48 -0.52 -0.45 -0.37 -0.47 -0.38 -0.47 -0.45 -0.44 -0.37 -0.42 -0.25 -0.47 -0.39 -0.38 -0.38 -0.50 -0.40 -0.53 -0.29 -0.34 -0.37 -0.36 -0.26 -0.34 -0.42 -0.49 -0.29 -0.38 -0.35 -0.45 -0.28 -0.34 -0.30 -0.43 -0.30 -0.49 -0.40 -0.38 -0.31 -0.53 -0.47 -0.57 -0.54 -0.59 -0.55 -0.87 -0.46 -0.51 -0.37 -0.62 -0.37 -0.59 -0.50 -0.65 -0.60 -0.62 -0.68 -0.56 -0.43 -0.41
frequency = 2