SSM hurricane Notebooks update#804
Conversation
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
|
@Dekermanjian should we solve the conflicts or close this PR? |
|
@aloctavodia my apologies. For some reason I thought this PR was merged a long time ago. I will resolve the conflicts. |
rebased main to resolve conflicts
2. Improved 24-hour ahead forecasts of simplest model 3. Added a section and model with how to add exogenous variables to the model
2. cleaned up typos and figure sizes/zoom
2. Added a non-linear B-Splines section
2. Added closing remarks section 3. Updated references bib file
2. Restructured function docstrings to comply with numpy's guidestyle 3. Added ODE derivation of Newtonian equations of motion 4. Added legend to origins scatterplot 5. Fixed issue with passing in future covariates into exogenous model when forecasting 6. Added model evaluations and comparisons 7. Fixed issue with diffuse initialization of P0
2. updated generate forecast utility 3. updated how mode=JAX is specified 4. renamed simple model to Newtonian model 5. added explanations to exogenous data duplicating and mirroring 6. added note on how the newtonian model differs from typical parameteric models
…ans and covariances. Updated plotly setting to automatically select renderer
…state means and covariances. Updated plotly setting to automatically select renderer" This reverts commit 07355d2. Revert using conditional posterior to using save_kalman_filter_outputs_in_idata
… with A and sometimes with T to only being referred to using T
…lter_outputs method
Dekermanjian
force-pushed
the
ssm_hurricane
branch
from
61e021c to
fa68d2c
Compare
|
View / edit / reply to this conversation on ReviewNB aloctavodia commented on 2026-02-14T10:01:13Z Can you also update to use "new arviz"? You need to use
import arviz.preview as az And then update some of the plot for instance instead of az.plot_trace use az.plot_trace_dist or az.plot_rank_dist. More information here https://python.arviz.org/en/latest/user_guide/migration_guide.html aloctavodia commented on 2026-02-15T06:45:00Z Use arviz style "arviz-variat" to match the style of other notebooks.
It's possible to use plotly as the plotting backend of arviz. To can set it globablly using rcParams. In order for the style to also work with plotly I think you need to set the style after importing plotly. |
|
Great, thanks, this is ready to merge, but I would like to know if you could also update to use the new arviz? If not, we can update in a different PR. |
Thank you @aloctavodia! Yes, I will update it to the new arviz today. |
|
Use arviz style "arviz-variat" to match the style of other notebooks.
It's possible to use plotly as the plotting backend of arviz. To can set it globablly using rcParams. In order for the style to also work with plotly I think you need to set the style after importing plotly. View entire conversation on ReviewNB |
2 participants
updated notebook to utilize the newly added
sample_filter_outputsmethod.Previously we were pulling out the filter outputs by passing
save_kalman_filter_outputs_in_idatato thebuild_statespacemethod. This will be deprecated in the future and we have added a utility methodsample_filter_outputsto allow users to sample the KF outputs.📚 Documentation preview 📚: https://pymc-examples--804.org.readthedocs.build/en/804/