move data-collection functions

src/surfaces/data_collector/config.py

@@ -1,3 +1,8 @@
+# Author: Simon Blanke
+# Email: simon.blanke@yahoo.com
+# License: MIT License
+
+
 import os
 
 here_path = os.path.dirname(os.path.realpath(__file__))

src/surfaces/data_collector/surfaces_data_collector.py

@@ -1,4 +1,16 @@
+# Author: Simon Blanke
+# Email: simon.blanke@yahoo.com
+# License: MIT License
+
+
+import numpy as np
+import pandas as pd
+from functools import reduce
+
 from search_data_collector import SqlSearchData
+from hyperactive import Hyperactive
+from hyperactive.optimizers import GridSearchOptimizer as HyperactiveGridSearchOptimizer
+from gradient_free_optimizers import GridSearchOptimizer
 
 from .config import default_search_data_path
 
@@ -8,3 +20,82 @@ def __init__(self, path=None) -> None:
         if path is None:
             path = default_search_data_path
         super().__init__(path, func2str=True)
+
+    def _init_search_data(self, objective_function, search_space):
+        self.para_names = [key for key in list(search_space.keys())]
+        self.search_data_length = 0
+
+        dim_sizes_list = [len(array) for array in search_space.values()]
+        self.search_space_size = reduce((lambda x, y: x * y), dim_sizes_list)
+
+        search_data_cols = self.para_names + ["score"]
+        self.search_data = pd.DataFrame([], columns=search_data_cols)
+
+    def _array_search_space(self, objective_function, search_space):
+        while self.search_data_length < self.search_space_size:
+            print("\n ------------ search_space_size", self.search_space_size)
+            opt = GridSearchOptimizer(
+                search_space,
+                direction="orthogonal",
+                initialize={},
+            )
+            opt.search(
+                objective_function,
+                n_iter=int(self.search_space_size * 1),
+                verbosity=["progress_bar"],
+            )
+
+            self.search_data = pd.concat(
+                [self.search_data, opt.search_data],
+                ignore_index=True,
+            )
+
+            self.search_data = self.search_data.drop_duplicates(subset=self.para_names)
+            self.search_data_length = len(self.search_data)
+            print(
+                "\n ------------ self.search_data_length", self.search_data_length, "\n"
+            )
+
+    def _list_search_space(self, objective_function, search_space):
+        while self.search_data_length < self.search_space_size:
+            print("\n ------------ search_space_size", self.search_space_size)
+
+            hyper = Hyperactive(verbosity=["progress_bar"])
+            hyper.add_search(
+                objective_function,
+                search_space,
+                initialize={},
+                n_iter=self.search_space_size,
+                optimizer=HyperactiveGridSearchOptimizer(direction="orthogonal"),
+                memory_warm_start=self.search_data,
+            )
+            hyper.run()
+
+            self.search_data = pd.concat(
+                [self.search_data, hyper.search_data(objective_function)],
+                ignore_index=True,
+            )
+
+            self.search_data = self.search_data.drop_duplicates(subset=self.para_names)
+            self.search_data_length = len(self.search_data)
+            print(
+                "\n ------------ self.search_data_length", self.search_data_length, "\n"
+            )
+
+    def collect(
+        self,
+        objective_function,
+        search_space,
+        table=None,
+        if_exists="append",
+    ):
+        if table is None:
+            table = objective_function.__name__
+
+        self._init_search_data(objective_function, search_space)
+        if isinstance(search_space[self.para_names[0]], np.ndarray):
+            self._array_search_space(objective_function, search_space)
+        else:
+            self._list_search_space(objective_function, search_space)
+
+        self.save(table, self.search_data, if_exists)

src/surfaces/test_functions/_base_test_function.py

@@ -4,8 +4,6 @@
 
 import time
 
-from ..data_collector import SurfacesDataCollector
-
 
 class BaseTestFunction:
     explanation = """ """
@@ -18,8 +16,6 @@ class BaseTestFunction:
     pure_objective_function: callable
 
     def __init__(self):
-        self.sql_data = SurfacesDataCollector()
-
         self.create_objective_function()
 
     def create_objective_function(self):

src/surfaces/test_functions/machine_learning/base_machine_learning_function.py

@@ -28,42 +28,6 @@ def __init__(self, metric=None, sleep=0, load_search_data=False):
 
         # self.objective_function.__func__.__name__ = self.__name__
 
-    def _collect_data(self, table=None, if_exists="append"):
-        if table is None:
-            table = self.__name__
-
-        search_space = self.search_space()
-        search_data_cols = self.para_names + ["score"]
-        search_data = pd.DataFrame([], columns=search_data_cols)
-        search_data_length = 0
-
-        dim_sizes_list = [len(array) for array in search_space.values()]
-        search_space_size = reduce((lambda x, y: x * y), dim_sizes_list)
-
-        while search_data_length < search_space_size:
-            print("\n ------------ search_space_size", search_space_size)
-
-            hyper = Hyperactive(verbosity=["progress_bar"])
-            hyper.add_search(
-                self.objective_function,
-                search_space,
-                initialize={},
-                n_iter=search_space_size,
-                optimizer=GridSearchOptimizer(direction="orthogonal"),
-                memory_warm_start=search_data,
-            )
-            hyper.run()
-
-            search_data = pd.concat(
-                [search_data, hyper.search_data(self.objective_function)],
-                ignore_index=True,
-            )
-
-            search_data = search_data.drop_duplicates(subset=self.para_names)
-            search_data_length = len(search_data)
-            print("\n ------------ search_data_length", search_data_length, "\n")
-        self.sql_data.save(self.__name__, search_data, if_exists)
-
     def objective_function_loaded(self, params):
         try:
             parameter_d = params.para_dict

src/surfaces/test_functions/mathematical/_base_objective_function.py

@@ -45,7 +45,10 @@ def add_dim(search_space_: dict, dim: int, min, max):
                 values = list(values)
             search_space_[dim_str] = values
 
-        if isinstance(min, list) and isinstance(max, list) and len(min) == len(max):
+        if isinstance(min, list) and isinstance(max, list):
+            if len(min) != len(max) or len(min) != self.n_dim:
+                raise ValueError
+
             for dim, (min_, max_) in enumerate(zip(min, max)):
                 add_dim(search_space_, dim, min_, max_)
         else:
@@ -63,45 +66,3 @@ def search_space_from_blank(search_space_blank, value_types):
                 dim_values = list(dim_values)
             search_space[para_names] = dim_values
         return search_space
-
-    def search_data(self):
-        self.search_space = self.search_space(value_types="array")
-
-        para_names = list(self.search_space.keys())
-        search_data_cols = para_names + ["score"]
-        search_data = pd.DataFrame([], columns=search_data_cols)
-        search_data_length = 0
-
-        dim_sizes_list = [len(array) for array in self.search_space.values()]
-        search_space_size = reduce((lambda x, y: x * y), dim_sizes_list)
-
-        self.create_objective_function()
-
-        while search_data_length < search_space_size:
-            print("\n ------------ search_space_size", search_space_size)
-            opt = GridSearchOptimizer(
-                self.search_space,
-                direction="orthogonal",
-                initialize={},
-            )
-            opt.search(
-                self.pure_objective_function,
-                n_iter=int(search_space_size * 1),
-                verbosity=["progress_bar"],
-            )
-
-            search_data = pd.concat(
-                [search_data, opt.search_data],
-                ignore_index=True,
-            )
-
-            search_data = search_data.drop_duplicates(subset=para_names)
-            search_data_length = len(search_data)
-            print("\n ------------ search_data_length", search_data_length, "\n")
-
-        return search_data
-
-    def _collect_data(self, table=None, if_exists="append"):
-        if table is None:
-            table = self.__name__
-        self.sql_data.save(table, self.search_data(), if_exists)