fix user guide

Author: SimonBlanke

docs/source/user_guide.rst

@@ -9,8 +9,11 @@ algorithm selection, and advanced features for efficient black-box optimization.
 
 .. tip::
 
-   New to GFO? Start with :ref:`search_spaces` to understand how to define
-   parameter spaces, then explore :ref:`optimizers` to choose your algorithm.
+   New to GFO? Follow this path through the guide:
+
+   1. :doc:`user_guide/search_spaces` -- learn how to define parameter ranges
+   2. :doc:`user_guide/objective_functions` -- write functions to optimize
+   3. :doc:`user_guide/optimizer_selection` -- choose the right algorithm for your problem
 
 ----
 
@@ -191,10 +194,9 @@ GFO provides 22 optimization algorithms organized into four categories:
       :link-type: doc
       :class-card: sd-border-start sd-border-warning
 
-      **4 algorithms** that learn from evaluations
+      **3 algorithms** that learn from evaluations
 
-      Bayesian Optimization, TPE, Forest Optimizer,
-      and Ensemble methods.
+      Bayesian Optimization, TPE, and Forest Optimizer.
 
 ----
 

docs/source/user_guide/optimizer_selection.rst

@@ -40,7 +40,7 @@ where to search next.
    .. grid-item-card:: Sequential Model-Based
       :class-card: gfo-compact gfo-smbo
 
-      Bayesian, TPE, Forest, Ensemble.
+      Bayesian, TPE, Forest.
       Best for expensive objective functions.
 
 

docs/source/user_guide/optimizers/index.rst

@@ -83,9 +83,6 @@ Algorithm Overview
     * - :doc:`Forest Optimizer <smbo/forest>`
       - SMBO
       - Large search spaces with discrete parameters
-    * - :doc:`Ensemble Optimizer <smbo/ensemble>`
-      - SMBO
-      - Robust surrogate modeling
 
 
 Categories
@@ -135,7 +132,7 @@ Categories
         Build a surrogate model of the objective function to predict promising
         regions. Ideal for expensive evaluations.
 
-        **Algorithms:** Bayesian Optimization, TPE, Forest Optimizer, Ensemble
+        **Algorithms:** Bayesian Optimization, TPE, Forest Optimizer
 
 
 How to Choose

docs/source/user_guide/optimizers/smbo/bayesian.rst

@@ -247,4 +247,3 @@ Related Algorithms
 
 - :doc:`tpe` - Density-based, handles categoricals well
 - :doc:`forest` - Tree-based, scales better
-- :doc:`ensemble` - Combines multiple models

docs/source/user_guide/optimizers/smbo/ensemble.rst

@@ -234,4 +234,3 @@ Related Algorithms
 
 - :doc:`bayesian` - GP-based, better uncertainty
 - :doc:`tpe` - Density-based, similar use cases
-- :doc:`ensemble` - Combines multiple models

docs/source/user_guide/optimizers/smbo/forest.rst

@@ -22,8 +22,6 @@ Overview
       - Tree-structured Parzen Estimators using density estimation.
     * - :doc:`forest`
       - Random Forest or Extra Trees as surrogate model.
-    * - :doc:`ensemble`
-      - Combines multiple surrogate models for robustness.
 
 
 How SMBO Works
@@ -129,10 +127,6 @@ Algorithm Comparison
       - Random Forest
       - Variance across trees
       - Large discrete spaces
-    * - Ensemble
-      - Multiple models
-      - Combined
-      - Robust predictions
 
 
 Surrogate Model Comparison
@@ -142,9 +136,9 @@ Surrogate Model Comparison
     :alt: SMBO surrogate model comparison
     :align: center
 
-    The four SMBO approaches use different surrogate models: GP with
-    uncertainty bands, TPE with good/bad density separation, Forest with
-    tree ensemble variance, and Ensemble combining multiple model types.
+    The three SMBO approaches use different surrogate models: GP with
+    uncertainty bands, TPE with good/bad density separation, and Forest with
+    tree ensemble variance.
 
 
 Visualization
@@ -226,4 +220,3 @@ Algorithms
     bayesian
     tpe
     forest
-    ensemble

docs/source/user_guide/optimizers/smbo/index.rst

@@ -222,4 +222,3 @@ Related Algorithms
 
 - :doc:`bayesian` - GP-based, better for continuous
 - :doc:`forest` - Tree-based surrogate
-- :doc:`ensemble` - Combines multiple models