Keyboard shortcuts

Press ← or → to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Statscraft 0.15.0, release v0.15.0

SEASON

Seasonal decomposition by ratios (or differences) to a moving average.

Syntax

SEASON [VARIABLES=] series [/MODEL=MULTIPLICATIVE | ADDITIVE] [/MA=EQUAL | CENTERED]
       [/PERIOD=n] [/APPLY].
The period comes from PERIOD, else TSET PERIOD, else the DATE variables; the series needs
at least four full periods and no missing value. A moving average over one period gives the
trend: for an even period CENTERED (the default) spans period + 1 cases with half weight at
both ends; EQUAL weighs the period cases equally, from period/2 cases before a case to
period/2 - 1 after it. An odd period always uses EQUAL. The multiplicative model divides the series by it (times 100), the
additive one subtracts it. For each position in the period the mean of these values without
the largest and the smallest is the raw factor; the factors are scaled to average 100
(multiplicative) or centered on 0 (additive). The adjusted series SAS is the series divided
by (or less) the factor, the trend-cycle STC a 1-2-3-2-1 weighted average of SAS (3-term
at the second and next-to-last case, extrapolated at the ends) and the irregular part ERR
SAS over (or less) STC. Tables: the model, the cases with every component (left out under
TSET PRINT=BRIEF) and the seasonal factors. TSET NEWVAR saves ERR_n, SAS_n, SAF_n and STC_n
per series; saved multiplicative factors are ratios, so SAS = series / SAF.
APPLY without a name repeats the previous SEASON.

Example

SEASON VARIABLES=heating_kwh /MODEL=ADDITIVE /PERIOD=12.

statscraft help SEASON prints the same text in the terminal. All commands