---
title: Signal Decomposition
slug: litmusedge/signal-decomposition
docTags: 
createdAt: 2024-08-06T18:58:20.949Z
---

The Signal Decomposition processor separates a time series into trend, seasonality, and residual components.

## Why use Signal Decomposition

Use this processor when you need to isolate a long-term trend or a repeating seasonal pattern from a signal's noise, for example, separating a machine's underlying wear trend from its normal cyclical variation.

## How it works

The processor calculates trend using linear regression and seasonality using naive differencing, then combines them with the residual (whatever's left over) according to the model you choose in **Model Type**:

- **Additive model**: `Signal = Trend + Seasonality + Residual`
- **Multiplicative model**: `Signal = Trend × Seasonality × Residual`

Use **Periodicity** to set the frequency at which the seasonal component is expected to repeat, and **Window Size** to set how many values the processor considers for each calculation.

## Parameters

| **Parameter** | **Details**                                                                               |
| ------------- | ----------------------------------------------------------------------------------------- |
| Window Size   | How many values are included in each periodic interval for the decomposition calculation. |
| Model Type    | The decomposition model to use: additive or multiplicative.                               |
| Periodicity   | The frequency at which the seasonal component is expected to repeat in the data.          |

## Related topics

- [Use the Signal Decomposition Function](docId\:CIVP10gpZ5rUGlReLvGBA)