tsdecomp
Splits a series into its trend, its seasonal pattern and what is left.
What we have
- additive and
multiplicativedecomposition, withgenerate() - checked against R's
decompose()
Coming later
arima,dfullerand forecasting, which are not built- it is not a Stata command, so there is no Stata output to compare with
You may shorten it to tsdeco.
How you write it
tsdecomp var, period(#) [multiplicative] [generate(stub)]
Square brackets mean optional. A vertical bar means choose one.
What it does
Sales rise every December. Is this December good? tsdecomp sales, period(12) separates the long movement (trend), the repeating within-year shape (seasonal) and the leftover (remainder), each as its own column, so you can answer that.
multiplicative is for series whose swings grow with their level — if the December bump is "20% up" rather than "5,000 up", use it.
Watch out. It needs every period present and at least two full cycles. It is not a Stata command — Stata has no
tsdecomp — so nothing here corresponds to Stata output. The numbers match R's decompose().Try it
Every example below was run after sysuse vira.sample04, clear.
. tsset month
Time variable: month, 1 to 96
Delta: 1 unit
The same thing in pandas and in R
pandas
# tsset: sorted by the time variable; L. and D. count in time, not rows
R
# no R form for `tsset` yet -- see docs/r-equivalents.md
. tsdecomp revenue, period(12)
Classical additive decomposition of revenue, period 12
season | index
---------+--------------
1 | -335.9028
2 | -282.1778
3 | -244.9290
4 | -61.2034
5 | 123.3740
6 | 267.2853
7 | 328.6198
8 | 290.7656
9 | 251.1531
10 | 35.0293
11 | -96.2475
12 | -275.7665
created: revenue_trend, revenue_seasonal, revenue_resid
(the trend and residual are missing for the first and last 6 periods)
The same thing in pandas and in R
pandas
from statsmodels.tsa.seasonal import seasonal_decompose parts = seasonal_decompose(df["revenue"], model="additive", period=12, two_sided=True) df["revenue_trend"] = parts.trend df["revenue_seasonal"] = parts.seasonal df["revenue_resid"] = parts.resid
R
series <- ts(df$revenue, frequency = 12) decomposed <- decompose(series, type = "additive") # This R was run in a real R (4.6.1) and gave the same numbers.
. summarize revenue_trend revenue_seasonal revenue_resid
Variable | Obs Mean Std. dev. Min Max -------------+------------------------------------------------------------ revenue_tre~ | 84 1151.523 178.1546 879.4167 1481.312 revenue_sea~ | 96 1.7171e-14 240.1762 -335.9028 328.6198 revenue_res~ | 84 -3.141815 122.9338 -400.5531 273.1552
The same thing in pandas and in R
pandas
df[['revenue_trend', 'revenue_seasonal', 'revenue_resid']].describe()
R
library(dplyr)
df %>% summarise(revenue_trend_n = sum(!is.na(revenue_trend)),
revenue_trend_mean = mean(revenue_trend, na.rm = TRUE),
revenue_trend_sd = sd(revenue_trend, na.rm = TRUE),
revenue_trend_min = min(revenue_trend, na.rm = TRUE),
revenue_trend_max = max(revenue_trend, na.rm = TRUE),
revenue_seasonal_n = sum(!is.na(revenue_seasonal)),
revenue_seasonal_mean = mean(revenue_seasonal, na.rm = TRUE),
revenue_seasonal_sd = sd(revenue_seasonal, na.rm = TRUE),
revenue_seasonal_min = min(revenue_seasonal, na.rm = TRUE),
revenue_seasonal_max = max(revenue_seasonal, na.rm = TRUE),
revenue_resid_n = sum(!is.na(revenue_resid)),
revenue_resid_mean = mean(revenue_resid, na.rm = TRUE),
revenue_resid_sd = sd(revenue_resid, na.rm = TRUE),
revenue_resid_min = min(revenue_resid, na.rm = TRUE),
revenue_resid_max = max(revenue_resid, na.rm = TRUE))
# This R was run in a real R (4.6.1) and gave the same numbers.
The full built-in help (help tsdecomp)
tsset timevar [, delta(#)] | tsset panelvar timevar
L.x L2.x F.x D.x D2.x S12.x in any expression
tsline var [var ...]
tssmooth ma newvar = var, window(#l #c #f)
tsdecomp var, period(#) [multiplicative] [generate(stub)]
. tsset month
. generate growth = D.sales / L.sales
. generate yoy = S12.sales / L12.sales
. tsdecomp sales, period(12)
. tsline sales sales_trend
The operators count in units of the time variable, NOT in rows. L.x at
month 5 is x at month 4, and it is missing if there is no month 4 -- even
when the row above is month 3. A plain row shift gets that wrong quietly,
and most real series have a missing month somewhere. In a panel each unit
is lagged on its own.
L. / F. a lag / a lead; L2. is two periods
D. a difference: x - L.x; D2. is a difference of differences
S12. a seasonal difference: x - L12.x
tssmooth ma window(2 1 2) is a 5-term centred average. The edges, where
the window does not fit, are missing
tsdecomp classical decomposition into trend, seasonal and remainder
(created as var_trend, var_seasonal, var_resid). Needs every
period present and two full cycles. Not a Stata command
Not built: date formats (the time variable is a plain number: 1, 2, 3 or
2019, 2020), arima, Holt-Winters, exponential smoothing, dfuller, forecasting.
The decomposition and moving average are checked against R's decompose() and
stats::filter(); the moving average's edge handling has not been compared
with Stata.