Time series data is a function of the sum of its components. The seasonal variation is relatively constant over time
The additive model is:
Y = T + C + S + e
1)
Compute the Trend-Cycle.
For quarterly data use a 4-moving-average (see multiplicative decomposition).
2)
Compute the seasonal component.
Y - Trend-Cycle (TC) = S + e
We need to smooth the seasonal data (see multiplicative decomposition).
3)
From the Trend-Cycle we infer the trend (T) with trendcurves (line, parabola, ..).
4)
The errors are:
Y - T - C - S = e