fixed broken reference in closing remarks

Author: Dekermanjian

examples/case_studies/ssm_hurricane_tracking.ipynb

@@ -38704,7 +38704,7 @@
     "# Closing Remarks\n",
     "In this case study we looked at how we can track a hurricane in two-dimensional space using a state space representation of Newtonian kinematics. We proceeded to expand on the pure Newtonian model and added exogenous variables that may hold information pertintent to the Hurricane's track. We then expanded our model by modeling our variables as smooth functions using cubic B-splines. \n",
     "\n",
-    "Throughout, the case study we have been evaluating our 24-hour forecasts and our overall mean error is smallest with our first Newtonian model. Below you will find the errors from all three models plotting against one another. It seems that (as expected) the exogenous information we included in the exogenous model was not informative with respect to the hurricances' trajectory. However, it is worth noting that in the period (around 40 through 56) where the hurricane manuevers we obtain less spikes in error in that section with our cubic B-spline model. This implies that the model could benefit from some non-linear specification to handle the angular acceleration. Hopefully, someday the `StateSpace` module in `pymc-extras` may support non-linear state space specifications with either the Extended Kalman Filter or with the Unscented Kalman Filter. Until then you can learn more about how to build your own custom state space models with the `StateSpace` module here {ref}`Making a Custom Statespace Model <>`."
+    "Throughout, the case study we have been evaluating our 24-hour forecasts and our overall mean error is smallest with our first Newtonian model. Below you will find the errors from all three models plotting against one another. It seems that (as expected) the exogenous information we included in the exogenous model was not informative with respect to the hurricances' trajectory. However, it is worth noting that in the period (around 40 through 56) where the hurricane manuevers we obtain less spikes in error in that section with our cubic B-spline model. This implies that the model could benefit from some non-linear specification to handle the angular acceleration. Hopefully, someday the `StateSpace` module in `pymc-extras` may support non-linear state space specifications with either the Extended Kalman Filter or with the Unscented Kalman Filter. Until then you can learn more about how to build your own custom state space models with the `StateSpace` module on the `pymc-extras` official GitHub repository."
    ]
   },
   {

examples/case_studies/ssm_hurricane_tracking.myst.md

@@ -1892,7 +1892,7 @@ plot_model_evaluations(
 # Closing Remarks
 In this case study we looked at how we can track a hurricane in two-dimensional space using a state space representation of Newtonian kinematics. We proceeded to expand on the pure Newtonian model and added exogenous variables that may hold information pertintent to the Hurricane's track. We then expanded our model by modeling our variables as smooth functions using cubic B-splines. 
 
-Throughout, the case study we have been evaluating our 24-hour forecasts and our overall mean error is smallest with our first Newtonian model. Below you will find the errors from all three models plotting against one another. It seems that (as expected) the exogenous information we included in the exogenous model was not informative with respect to the hurricances' trajectory. However, it is worth noting that in the period (around 40 through 56) where the hurricane manuevers we obtain less spikes in error in that section with our cubic B-spline model. This implies that the model could benefit from some non-linear specification to handle the angular acceleration. Hopefully, someday the `StateSpace` module in `pymc-extras` may support non-linear state space specifications with either the Extended Kalman Filter or with the Unscented Kalman Filter. Until then you can learn more about how to build your own custom state space models with the `StateSpace` module here {ref}`Making a Custom Statespace Model <>`.
+Throughout, the case study we have been evaluating our 24-hour forecasts and our overall mean error is smallest with our first Newtonian model. Below you will find the errors from all three models plotting against one another. It seems that (as expected) the exogenous information we included in the exogenous model was not informative with respect to the hurricances' trajectory. However, it is worth noting that in the period (around 40 through 56) where the hurricane manuevers we obtain less spikes in error in that section with our cubic B-spline model. This implies that the model could benefit from some non-linear specification to handle the angular acceleration. Hopefully, someday the `StateSpace` module in `pymc-extras` may support non-linear state space specifications with either the Extended Kalman Filter or with the Unscented Kalman Filter. Until then you can learn more about how to build your own custom state space models with the `StateSpace` module on the `pymc-extras` official GitHub repository.
 
 ```{code-cell} ipython3
 fig = go.Figure()