+ reran again to ensure cell execution order even for markdown cells

Author: jonsedar

examples/generalized_linear_models/GLM-ordinal-features.ipynb

@@ -2143,9 +2143,9 @@
     "+ `chi_d450`: Non-linear response throughout the range\n",
     "+ `nu_d450`: The non-linear effect `beta * chi.csum()` is clear, in particular `c0` is far from the trend of `c1, c2, c3`\n",
     "\n",
-    "Note in particular that the posterior distribution of `chi_d450 = c4` (and thus `nu_d450 = c4`) is almost exactly the \n",
+    "Note in particular that the posterior distribution of `chi_d450 = c4` is almost exactly the \n",
     "same value as for its prior, because it hasn't been evidenced in the dataset. The constraint of the Dirichlet has in\n",
-    "turn scaled the values for `c0` to `c3` and the scale of `beta_450`. \n",
+    "turn scaled the values for `chi_d450 = c0 to c3`, the scale of `beta_450`, and downstream effects on `nu_d450 = c4`.\n",
     "\n",
     "For comparison you can try the inferior alternative by setting `COORDS['d450_nm']=list(df[d450].cat.categories)` in the\n",
     "model spec in Section 2.1 and re-running and seeing what happens"

examples/generalized_linear_models/GLM-ordinal-features.myst.md

@@ -1454,9 +1454,9 @@ Interesting pattern:
 + `chi_d450`: Non-linear response throughout the range
 + `nu_d450`: The non-linear effect `beta * chi.csum()` is clear, in particular `c0` is far from the trend of `c1, c2, c3`
 
-Note in particular that the posterior distribution of `chi_d450 = c4` (and thus `nu_d450 = c4`) is almost exactly the 
+Note in particular that the posterior distribution of `chi_d450 = c4` is almost exactly the 
 same value as for its prior, because it hasn't been evidenced in the dataset. The constraint of the Dirichlet has in
-turn scaled the values for `c0` to `c3` and the scale of `beta_450`. 
+turn scaled the values for `chi_d450 = c0 to c3`, the scale of `beta_450`, and downstream effects on `nu_d450 = c4`.
 
 For comparison you can try the inferior alternative by setting `COORDS['d450_nm']=list(df[d450].cat.categories)` in the
 model spec in Section 2.1 and re-running and seeing what happens