-
Notifications
You must be signed in to change notification settings - Fork 2.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
ENH: state space: compute smoothed state autocovariance matrices for arbitrary lags #6579
Conversation
Codecov Report
@@ Coverage Diff @@
## master #6579 +/- ##
==========================================
+ Coverage 85.28% 85.31% +0.02%
==========================================
Files 646 646
Lines 103651 103922 +271
Branches 11270 11311 +41
==========================================
+ Hits 88401 88657 +256
- Misses 12797 12806 +9
- Partials 2453 2459 +6
Continue to review full report at Codecov.
|
94ad3d4
to
61a94bc
Compare
Squashed. |
Also, this is still a semi-private method, since this PR only puts a new method in |
4e68c1a
to
f438e9d
Compare
f438e9d
to
37272a4
Compare
We already compute the
lag=1
smoothed autocovariance matrix for the state vector (i.e. Cov(t, t-1)) in the Cython recursions (because it is required for the EM algorithm applied to the transition equation, and so it's best if it's pretty fast).This PR adds a new
smoothed_state_autocovariance
method to theSmootherResults
class to allow computing autocovariances with anylag = 0, +-1, +-2, ...
. One application of this is computing the "news" for nowcasting models, described in Banbura and Modugo (2014). Not in Cython since as far as I know, this is just for postestimation results, and won't be in an inner loop.