Decomposition results
Interpretation:
Component plots
Residual diagnostics
Seasonal indices
| Season | Index | Average observed value | Observations |
|---|
Detailed decomposition table
| Date | Observed | Trend | Seasonal | Residual | Adjusted | Reconstructed | Outlier |
|---|
Model comparison
| Method | Model | Period | Residual RMSE | MAE | Trend strength | Seasonal strength | Recommendation |
|---|---|---|---|---|---|---|---|
| Run a comparison after calculating the decomposition. | |||||||
Formula used
Additive model:
Observed = Trend + Seasonal + Residual
Multiplicative model:
Observed = Trend × Seasonal × Residual
Seasonally adjusted series:
Additive: Adjusted = Observed − Seasonal
Multiplicative: Adjusted = Observed ÷ Seasonal
Trend strength ≈ max(0, 1 − Var(Residual) ÷ Var(Trend + Residual))
Seasonal strength ≈ max(0, 1 − Var(Residual) ÷ Var(Seasonal + Residual))
How to use
- Paste or upload chronological dates and numeric values.
- Select a decomposition method, model, frequency, and seasonal period.
- Choose cleaning, transformation, trend, and chart settings.
- Calculate the decomposition and inspect each component.
- Review residual diagnostics, seasonal indices, and model comparisons.
- Copy or export the detailed results and charts.
Example seasonal periods
| Data frequency | Common period | Example |
|---|---|---|
| Hourly | 24 or 168 | Daily or weekly electricity demand |
| Daily | 7, 30, or 365 | Weekly traffic or annual temperature cycles |
| Weekly | 52 | Annual retail demand |
| Monthly | 12 | Annual sales seasonality |
| Quarterly | 4 | Annual business cycles |
Frequently asked questions
What does seasonal decomposition calculate?
It separates observations into trend, seasonal, and residual components.
When should I use an additive model?
Use it when seasonal changes remain roughly constant in size.
When should I use a multiplicative model?
Use it when seasonal variation increases with the series level.
How many observations are required?
At least two complete seasonal cycles are strongly recommended.
What is STL decomposition?
STL uses repeated smoothing to estimate flexible trend and seasonality.
Why inspect residual autocorrelation?
Remaining autocorrelation suggests the decomposition missed useful structure.
Can the calculator handle missing values?
Yes. Several interpolation and row-removal methods are available.
What is a seasonally adjusted series?
It removes estimated seasonal effects while retaining trend and noise.
Can decomposition replace a forecasting model?
No. It supports diagnosis and preprocessing before formal forecasting.