Create search_space.md

docs/tutorial/search_space.md

@@ -0,0 +1,50 @@
+## Create the search space
+
+Since v0.5.0 the search space is created by defining:
+  - a <b>function</b> for the model
+  - a parameter <b>dictionary</b>
+
+
+The function receives 3 arguments:
+  - <b>para</b> : This defines what part of the model-function should be optimized
+  - <b>X</b> : Training features
+  - <b>y</b> : Training target
+
+ The function should return some kind of metric that will be <b>maximized</b> during the search.
+```python
+from sklearn.model_selection import cross_val_score
+from sklearn.ensemble import GradientBoostingClassifier
+
+def model(para, X, y):
+    model = GradientBoostingClassifier(
+        n_estimators=para["n_estimators"],
+        max_depth=para["max_depth"],
+    )
+    scores = cross_val_score(model, X, y, cv=3)
+
+    return scores.mean()
+```
+
+The search_config is a dictionary, that has the <b>model-function as a key</b> and its <b>values defines the search space</b> for this model. The search space is an additional dictionary that will be used in 'para' within the model-function.
+```python
+search_config = {
+    model: {
+        "n_estimators": range(10, 200, 10),
+        "max_depth": range(2, 12),
+        "min_samples_split": range(2, 12),
+    }
+}
+```
+
+This way of creating the search space has <b>multiple advantages</b>:
+  - No new syntax to learn. You can create the model as you are used to.
+  - It makes the usage of hyperactive very versatile, because you can define <b>any kind of function</b> and optimize it. This enables:
+    - The optimization of:
+      - complex machine-learning pipelines and ensembles
+      - deep neural network architecture
+    - The usage of <b>any machine learning framework</b> you like. The following are tested:
+      - Sklearn
+      - XGBoost
+      - LightGBM
+      - CatBoost
+      - Keras