Current implementation of fit_predict() and OxCGRT-parameters forecast

[Inglezos]

I noticed that with the current implementation of parameters forecast sometimes the predicted Rt value can become negative if delay is large. For example I tried hardcoded delay of 30 days instead of the recovery period and I run the estimator for Greece. The result was either to have a phase with 0 length in days or Rt to become small negative. Is it a bug or is this happening because the mechanism is not fully developed yet?

[lisphilar]

This is caused by unknown reasons and please create a bug-fix issue with the details.
Is rho value also negative?

[Inglezos]

Yes I think the rho values were also negative.

[Inglezos]

I created the bug issue #507.

[lisphilar]

Thank you. We do not have "close" feature "Discussions" beta version unfortunately.