Decompose Signal

Decomposes selected Time-Series or IoT signal into Trend, 2 Seasonal Components, and the remaining Residual.

Signal = T + S1 + S2 + R

[T] Trend Component: is calculated by fitting a regression model through the data with degree 2.
[S1] Seasonal Component 1: is calculated as the first major spike in auto-correlation.
[S2] Seasonal Component 2: is calculated as the first major spike in auto-correlation after the diferencing of the first seasonality.
[R] Residual: is what remains after trend and the two Seasonalities have been differenced.

The interactive displays shows the first 1000 records fpr the above outputs as well as the ACF plot for the detrended signal and both subsequent series after seasonality one and two are removed.

Required extensions:
KNIME Python Integration
(https://hub.knime.com/knime/extensions/org.knime.features.python2/latest)
KNIME Quick Forms
(https://hub.knime.com/knime/extensions/org.knime.features.js.quickforms/latest)
KNIME Math Expression (JEP)
(https://hub.knime.com/knime/extensions/org.knime.features.ext.jep/latest)
KNIME JavaScript Views
(https://hub.knime.com/knime/extensions/org.knime.features.js.views/latest)

Options

Signal Column
Select the column to decompose.
Correlation Cut-Off
Choose the minimum auto-correlation value for a seasonality to be considered.%%00010Value between 0 and 1.
Max Lags
Choose the maximum number of lags to calculate for ACF calculation when looking for seasonality.
Lag Step
Choose the step size between auto-correlation calculations when checking for seasonality.

Input Ports

Icon
Table containing signal column

Output Ports

Icon
Original table with Trend, Seasonality 1, Seasonality 2, and Residual columns added.
Icon
The Regression model representing the Singal's Trend

Nodes

Extensions

Links