search_space_example

Complexity

Total Complexity

7

Size/Duplication

Total Lines	65
Duplicated Lines	0 %

Importance

Changes

0

7 Functions

Rating	Name	Duplication	Size	Complexity
A	list2()	0	2	1
A	function_0()	0	3	1
A	function_2()	0	3	1
A	objective_function()	0	4	1
A	list1()	0	2	1
A	function_1()	0	3	1
A	list3()	0	2	1

"""
Hyperactive is very versatile, because it can handle not just numerical or
string variables in the search space, but also functions. If you want to
search for the best list, numpy array, dataframed or class you can put them into a
function that returns them as shown in the example below.

This enables many possibilities for more complex optimization applications.
Neural architecture search, feature engineering, ensemble optimization and many other applications are
only possible or much easier if you can put functions in the search space.
"""

from hyperactive import Hyperactive


def function_0():
    # do stuff in function0
    return


def function_1():
    # do stuff in function1
    return


def function_2():
    # do stuff in function2
    return


def list1():
    return [1, 0, 0]


def list2():
    return [0, 1, 0]


def list3():
    return [0, 0, 1]


# Hyperactive can handle python objects in the search space
search_space = {
    "int": list(range(1, 10)),
    "float": [0.1, 0.01, 0.001],
    "string": ["string1", "string2"],
    "function": [function_0, function_1, function_2],
    "list": [list1, list2, list3],
}


def objective_function(para):
    # score must be a number
    score = 1
    return score


hyper = Hyperactive()
hyper.add_search(objective_function, search_space, n_iter=20)
hyper.run()

search_data = hyper.search_data(objective_function)

print("\n Search Data: \n", search_data)