The tool supports interface of Foresight Analytics Platform 9 or earlier.
The Dickey-Fuller test is used to estimate the stability of a time series. Results of the test calculation for each tested variables are displayed on the appropriate panel:
Model Type. The test enables the user to work with equations of three types. The equation can be selected in the list:
Without Constant and Trend (describes the process of random walks).
With Constant.
With Constant and Trend.
Series Differentiation. It sets the equation differentiation order. The following options are available:
Source Series. Differentiation is not used.
Differentiated Series.
Twice Differentiated Series.
Autoregression Order. Autoregression order is set manually or using the value editor.
Use R Connection on Calculation. It determines whether to use R statistical package on calculating the Dickey-Fuller test. By default, the test is calculated by Foresight Analytics Platform methods. Select this checkbox to calculate the test by R methods.
NOTE. The option is available if Foresight Analytics Platform has R package connected. For details see the How to Set Up Integration with R? section.
The test is executed consecutively for each selected variable.
Test results are given in the table. Calculated ADF-statistics is compared with the N percent (N=5%, 1%) significance level, and the conclusion is made whether the series is stationary or non-stationary. This conclusion is also displayed in the table. The percentage significance level is set depending on the selected equation model and the length of the source series.
See also:
Descriptive Statistics | Calculation Method for The Dickey-Fuller Test