3
$\begingroup$

I have a time series signal (continuous values) and some events occurring. An event for me is characterized by the fact that it only has a single timestamp assigned to it and not a timespan.

Is there any correlation measure to calculate the relationship between the signal and the event? The use case would be, that I want to find out which events are responsible for an increase of the continuous signal, for example.

Thanks for any help.

$\endgroup$
  • 1
    $\begingroup$ You can do a point process regression. $\endgroup$ – Memming May 18 '15 at 15:29
1
$\begingroup$

This can be analyzed using Intervention Modeling. It is a regression with a causal variable where you have zeroes before the event and then a 1 from the event and forward. The t-value will tell you the magnitude of the impact.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.