Total Complexity | 7 |
Total Lines | 59 |
Duplicated Lines | 0 % |
Changes | 0 |
1 | """ |
||
2 | Hyperactive is very versatile, because of it can handle not just numerical or |
||
3 | string variables in the search space, but also functions. This enables many |
||
4 | possibilities for more complex optimization applications. Neural architecture search, |
||
5 | feature engineering, ensemble optimization and many other applications are |
||
6 | only possible or much easier, if you can put functions in the search space. |
||
7 | """ |
||
8 | |||
9 | from hyperactive import Hyperactive |
||
10 | |||
11 | |||
12 | def function_0(): |
||
13 | pass |
||
14 | |||
15 | |||
16 | def function_1(): |
||
17 | pass |
||
18 | |||
19 | |||
20 | def function_2(): |
||
21 | pass |
||
22 | |||
23 | |||
24 | def list1(): |
||
25 | return [1, 0, 0] |
||
26 | |||
27 | |||
28 | def list2(): |
||
29 | return [0, 1, 0] |
||
30 | |||
31 | |||
32 | def list3(): |
||
33 | return [0, 0, 1] |
||
34 | |||
35 | |||
36 | # Hyperactive can handle python objects in the search space |
||
37 | search_space = { |
||
38 | "int": list(range(1, 10)), |
||
39 | "float": [0.1, 0.01, 0.001], |
||
40 | "string": ["string1", "string2"], |
||
41 | "function": [function_0, function_1, function_2], |
||
42 | "list": [list1, list2, list3], |
||
43 | } |
||
44 | |||
45 | |||
46 | def objective_function(para): |
||
47 | # score must be a single number |
||
48 | score = 1 |
||
49 | return score |
||
50 | |||
51 | |||
52 | hyper = Hyperactive() |
||
53 | hyper.add_search(objective_function, search_space, n_iter=20) |
||
54 | hyper.run() |
||
55 | |||
56 | search_data = hyper.results(objective_function) |
||
57 | |||
58 | print("\n Search Data: \n", search_data) |
||
59 |