Merge pull request #7 from rodrigo-arenas/0.4.0dev

plots update

.travis.yml

@@ -1,6 +1,5 @@
 language: python
 python:
-  - 3.6
   - 3.7
   - 3.8
   - 3.9

README.rst

@@ -16,6 +16,7 @@
 .. |Docs| image:: https://readthedocs.org/projects/sklearn-genetic-opt/badge/?version=latest
 .. _Docs: https://sklearn-genetic-opt.readthedocs.io/en/latest/?badge=latest
 
+.. image:: /docs/logo.png
 
 Sklearn-genetic-opt
 ########
@@ -24,7 +25,19 @@ scikit-learn models hyperparameters tuning, using evolutionary algorithms.
 
 This is meant to be an alternative from popular methods inside scikit-learn such as Grid Search and Randomized Grid Search.
 
-Sklearn-genetic-opt uses evolutionary algorithms from the deap package to find the "best" set of hyperparameters that optimizes (max or min) the cross validation scores, it can be used for both regression and classification problems.
+Sklearn-genetic-opt uses evolutionary algorithms from the deap package to find the "best" set of hyperparameters that
+optimizes (max or min) the cross validation scores, it can be used for both regression and classification problems.
+
+Documentation is available `here <https://sklearn-genetic-opt.readthedocs.io/>`_
+
+Optimization progress in a regression problem:
+
+.. image:: https://github.com/rodrigo-arenas/Sklearn-genetic-opt/blob/master/demo/images/fitness.png?raw=True
+
+Sampled distribution oh hyperparameters
+
+.. image:: https://github.com/rodrigo-arenas/Sklearn-genetic-opt/blob/master/demo/images/density.png?raw=True
+
 
 Main Features:
 ########
@@ -49,7 +62,6 @@ Example
 .. code-block:: python
 
    from sklearn_genetic import GASearchCV
-   from sklearn_genetic.plots import plot_fitness_evolution, plot_search_space
    from sklearn_genetic.space import Continuous, Categorical, Integer
    from sklearn.ensemble import RandomForestClassifier
    from sklearn.model_selection import train_test_split, StratifiedKFold
@@ -91,14 +103,6 @@ Example
    y_predict_ga = evolved_estimator.predict(X_test)
    print(accuracy_score(y_test, y_predict_ga))
 
-   # Check the distribution of sampled hyperparameters
-   plot_search_space(evolved_estimator, features=['min_weight_fraction_leaf', 'max_depth', 'max_leaf_nodes', 'n_estimators'])
-   plt.show()
-
-   # See the evolution of the optimization per generation
-   plot_fitness_evolution(evolved_estimator)
-   plt.show()
-
    # Saved metadata for further analysis
    print("Stats achieved in each generation: ", evolved_estimator.history)
    print("Best k solutions: ", evolved_estimator.hof)
@@ -107,17 +111,9 @@ Example
 Results
 ^^^^^^^^^
 
-Sampled distribution
-
-.. image:: https://github.com/rodrigo-arenas/Sklearn-genetic-opt/blob/0.4.x/demo/images/density.png?raw=True
-
-Fitness evolution over generations
-
-.. image:: https://github.com/rodrigo-arenas/Sklearn-genetic-opt/blob/0.4.x/demo/images/fitness.png?raw=True
-
 Log controlled by verbosity
 
-.. image:: https://github.com/rodrigo-arenas/Sklearn-genetic-opt/blob/0.4.x/demo/images/log.JPG?raw=True
+.. image:: https://github.com/rodrigo-arenas/Sklearn-genetic-opt/blob/master/demo/images/log.JPG?raw=True
 
 
 Contributing

demo/Boson_Houses_decision_tree.py

@@ -30,7 +30,7 @@
     clf,
     cv=3,
     scoring="r2",
-    population_size=12,
+    population_size=20,
     generations=30,
     tournament_size=3,
     elitism=True,

dev-requirements.txt

@@ -11,4 +11,5 @@ black==21.5b2
 sphinx
 sphinx_gallery
 sphinx_rtd_theme
-sphinx-copybutton
+sphinx-copybutton
+numpydoc

docs/conf.py

@@ -64,6 +64,13 @@
 # so a file named "default.css" will overwrite the builtin "default.css".
 html_static_path = ["_static"]
 
+html_theme_options = {
+    "navigation_depth": 3,
+    "logo_only": True,
+}
+
+html_logo = "logo.png"
+
 master_doc = "index"
 
 # generate autosummary even if no references

docs/index.rst

@@ -3,10 +3,10 @@
    You can adapt this file completely to your liking, but it should at least
    contain the root `toctree` directive.
 
-Welcome to sklean-genetic-opt's documentation!
-==============================================
-
+sklean-genetic-opt
+==================
 scikit-learn models hyperparameters tuning, using evolutionary algorithms.
+##########################################################################
 
 This is meant to be an alternative from popular methods inside scikit-learn such as Grid Search and Randomized Grid Search.
 

docs/tutorials/callbacks.rst

@@ -70,7 +70,7 @@ using the 'fitness_min' value:
 
 ThresholdStopping
 -----------------
-IT stop the optimization if the current metric
+It stops the optimization if the current metric
 is greater or equals than the define threshold.
 
 For example, if we want to stop the optimization

docs/tutorials/custom_callback.rst

@@ -6,7 +6,7 @@ but you can make one of your own by defining a callable with
 certain methods.
 
 The callback must be a class that implements the ``__call__`` and
-``_check`` methods, the result of them must be a bool, ``True`` means
+``on_step`` methods, the result of them must be a bool, ``True`` means
 that the optimization must stop, ``False``, means it can continue.
 
 In this example, we are going to define a dummy callback that
@@ -17,7 +17,7 @@ The callback must have two parameters: `record` and `logbook`.
 Those are a dictionary and a deap's Logbook object respectively,
 with the current iteration metrics and all the past iterations metrics.
 You can choice which to use, but both must be parameters
-on the ``_check`` and ``__call__`` methods.
+on the ``on_step`` and ``__call__`` methods.
 
 So to check inside the logbook, we could define a function like this:
 
@@ -27,7 +27,7 @@ So to check inside the logbook, we could define a function like this:
     metric='fitness'
     threshold=0.8
 
-    def _check(record, logbook, threshold):
+    def on_step(record, logbook, threshold):
         # Not enough data points
         if len(logbook) <= N:
             return False
@@ -53,7 +53,7 @@ that will have all this parameters, so we can rewrite it like this:
            self.N = N
            self.metric = metric
 
-       def _check(self, record, logbook):
+       def on_step(self, record, logbook):
            # Not enough data points
            if len(logbook) <= self.N:
                return False
@@ -68,7 +68,7 @@ that will have all this parameters, so we can rewrite it like this:
            return False
 
        def __call__(self, record, logbook):
-           return self._check(record, logbook)
+           return self.on_step(record, logbook)
 
 
 So that is it, now you can initialize the DummyThreshold

ims.py

@@ -0,0 +1,43 @@
+import matplotlib.pyplot as plt
+from sklearn_genetic import GASearchCV
+from sklearn_genetic.space import Categorical, Integer, Continuous
+from sklearn.model_selection import train_test_split, StratifiedKFold
+from sklearn.ensemble import RandomForestClassifier
+from sklearn.datasets import load_digits
+from sklearn.metrics import accuracy_score
+
+data = load_digits()
+n_samples = len(data.images)
+X = data.images.reshape((n_samples, -1))
+y = data['target']
+X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)
+
+param_grid = {'min_weight_fraction_leaf': Continuous(0.01, 0.5, distribution='log-uniform'),
+              'bootstrap': Categorical([True, False]),
+              'max_depth': Integer(2, 30),
+              'max_leaf_nodes': Integer(2, 35),
+              'n_estimators': Integer(100, 300)}
+
+# The base classifier to tune
+clf = RandomForestClassifier()
+
+# Our cross-validation strategy (optional)
+cv = StratifiedKFold(n_splits=3, shuffle=True)
+
+# The main class from sklearn-genetic-opt
+evolved_estimator = GASearchCV(estimator=clf,
+                               cv=cv,
+                               generations=20,
+                               population_size=6,
+                               scoring='accuracy',
+                               param_grid=param_grid,
+                               algorithm='eaSimple',
+                               n_jobs=-1,
+                               verbose=True)
+
+# Train and optimize the estimator
+evolved_estimator.fit(X_train, y_train)
+
+from sklearn_genetic.plots import plot_search_space
+plot_search_space(evolved_estimator, features=['min_weight_fraction_leaf', 'max_depth', 'max_leaf_nodes', 'n_estimators'])
+plt.show()

sklearn_genetic/callbacks.py

@@ -77,7 +77,7 @@ def __init__(self, threshold, metric="fitness"):
         self.threshold = threshold
         self.metric = metric
 
-    def _check(self, record, logbook):
+    def on_step(self, record, logbook):
         """
         Parameters
         ----------
@@ -103,7 +103,7 @@ def _check(self, record, logbook):
             )
 
     def __call__(self, record=None, logbook=None):
-        return self._check(record, logbook)
+        return self.on_step(record, logbook)
 
 
 class ConsecutiveStopping:
@@ -126,7 +126,7 @@ def __init__(self, generations, metric="fitness"):
         self.generations = generations
         self.metric = metric
 
-    def _check(self, record=None, logbook=None):
+    def on_step(self, record=None, logbook=None):
         """
         Parameters
         ----------
@@ -156,7 +156,7 @@ def _check(self, record=None, logbook=None):
             raise ValueError("logbook parameter must be provided")
 
     def __call__(self, record=None, logbook=None):
-        return self._check(record, logbook)
+        return self.on_step(record, logbook)
 
 
 class DeltaThreshold:
@@ -179,7 +179,7 @@ def __init__(self, threshold, metric: str = "fitness"):
         self.threshold = threshold
         self.metric = metric
 
-    def _check(self, record=None, logbook=None):
+    def on_step(self, record=None, logbook=None):
         """
         Parameters
         ----------
@@ -210,4 +210,4 @@ def _check(self, record=None, logbook=None):
             raise ValueError("logbook parameter must be provided")
 
     def __call__(self, record=None, logbook=None):
-        return self._check(record, logbook)
+        return self.on_step(record, logbook)

sklearn_genetic/plots.py

@@ -2,7 +2,6 @@
 
 from .utils import logbook_to_pandas
 
-sns.set_style("whitegrid")
 
 """
 This module contains some useful function to explore the results of the optimization routines
@@ -21,6 +20,8 @@ def plot_fitness_evolution(estimator):
     Lines plot with the fitness value in each generation
 
     """
+    sns.set_style("white")
+
     fitness_history = estimator.history["fitness"]
 
     palette = sns.color_palette("rocket")
@@ -53,6 +54,8 @@ def plot_search_space(estimator, height=2, s=25, features: list = None):
     Pair plot of the used hyperparameters during the search
 
     """
+    sns.set_style("white")
+
     df = logbook_to_pandas(estimator.logbook)
     if features:
         stats = df[features]
@@ -61,7 +64,11 @@ def plot_search_space(estimator, height=2, s=25, features: list = None):
         stats = df[variables]
 
     g = sns.PairGrid(stats, diag_sharey=False, height=height)
-    g = g.map_upper(sns.scatterplot, s=s)
-    g = g.map_lower(sns.kdeplot, shade=True)
-    g = g.map_diag(sns.kdeplot, shade=True)
+    g = g.map_upper(sns.scatterplot, s=s, color="r", alpha=0.2)
+    g = g.map_lower(
+        sns.kdeplot,
+        shade=True,
+        cmap=sns.color_palette("ch:s=.25,rot=-.25", as_cmap=True),
+    )
+    g = g.map_diag(sns.kdeplot, shade=True, palette="crest", alpha=0.2, color="red")
     return g