Difference between revisions of "Orange: Granger Causality"
Jump to navigation
Jump to search
Onnowpurbo (talk | contribs) |
Onnowpurbo (talk | contribs) |
||
| Line 4: | Line 4: | ||
Test if one time series Granger-causes (i.e. can be an indicator of) another time series. | Test if one time series Granger-causes (i.e. can be an indicator of) another time series. | ||
| − | + | ==Input== | |
| − | + | Time series: Time series as output by As Timeseries widget. | |
This widgets performs a series of statistical tests to determine the series that cause other series so we can use the former to forecast the latter. | This widgets performs a series of statistical tests to determine the series that cause other series so we can use the former to forecast the latter. | ||
| Line 12: | Line 12: | ||
[[File:Granger-causality-stamped.png|center|200px|thumb]] | [[File:Granger-causality-stamped.png|center|200px|thumb]] | ||
| − | + | * Desired level of confidence. | |
| − | + | * Maximum lag to test to. | |
| − | + | * Runs the test. | |
| − | + | * Denotes the minimum lag at which one series can be said to cause another. In the first line of the example above, if we have the monthly unemployment rate time series for Austria, we can say something about unemployment rate in Hungary 10 months ahead. | |
| − | + | * The causing (antecedent) series. | |
| − | + | * The effect (consequent) series. | |
The time series that Granger-cause the series you are interested in are good candidates to have in the same VAR model. But careful, even if one series is said to Granger-cause another, this doesn’t mean there really exists a causal relationship. Mind your conclusions. | The time series that Granger-cause the series you are interested in are good candidates to have in the same VAR model. But careful, even if one series is said to Granger-cause another, this doesn’t mean there really exists a causal relationship. Mind your conclusions. | ||
Revision as of 06:25, 30 January 2020
Sumber: https://orange.biolab.si/widget-catalog/time-series/granger_causality/
Test if one time series Granger-causes (i.e. can be an indicator of) another time series.
Input
Time series: Time series as output by As Timeseries widget.
This widgets performs a series of statistical tests to determine the series that cause other series so we can use the former to forecast the latter.
- Desired level of confidence.
- Maximum lag to test to.
- Runs the test.
- Denotes the minimum lag at which one series can be said to cause another. In the first line of the example above, if we have the monthly unemployment rate time series for Austria, we can say something about unemployment rate in Hungary 10 months ahead.
- The causing (antecedent) series.
- The effect (consequent) series.
The time series that Granger-cause the series you are interested in are good candidates to have in the same VAR model. But careful, even if one series is said to Granger-cause another, this doesn’t mean there really exists a causal relationship. Mind your conclusions.