ARIMA Learner

Estimates parameters of an ARIMA time series model.

Options

Column containing univariate time series
Column containing the time series which should be used for fitting the model.
AR/I/MA order
Orders of appropriate components of the ARIMA model.
Allow non-zero mean
If selected, an estimate of the mean of the time series will be also estimated. Otherwise, the time series will be treated as if it had mean equal 0. Models with positive integration order are always treated as having zero mean. The way how the mean is estimated depends on the particular estimation methods. Conditional likelihood method estimates the mean together with other parameters while maximizing the loglikelihood function. Yule-Walker, Hannan-Rissanen, and Maximum Likelihood estimate the mean of the time series using sample mean, subtract it from the time series and then estimate other parameters using centered time series.
Estimation method
Method which should be used for estimating the parameters of the ARIMA model. Four methods are available:
  1. Conditional likelihood
    Recommended estimation method. Estimates parameters by maximizing conditional loglikelihood as given by Hamilton (1994) "Time Series Analysis", sec. 5.6.
  2. Yule - Walker
    Applicable only to pure AR processes (MA order must be equal 0). Estimates the parameters using Yule-Walker equations. For details see Brockwell & Davies (2003) "Introduction to Time Series and Forecasting", sec. 5.1.1.
  3. Maximum likelihood
    Maximum Likelihood estimation computed using Innovations Algorithm as described in Brockwell & Davies (2003) "Introduction to Time Series and Forecasting", sec. 5.2.

Input Ports

Icon
Table containing time series data.

Output Ports

Icon
Model to connect to a predictor node.

Views

This node has no views

Workflows

Links

Developers

You want to see the source code for this node? Click the following button and we’ll use our super-powers to find it for you.