|
1
|
|
|
"""Results handling for hyperparameter optimization. |
|
2
|
|
|
|
|
3
|
|
|
Author: Simon Blanke |
|
4
|
|
|
Email: [email protected] |
|
5
|
|
|
License: MIT License |
|
6
|
|
|
""" |
|
7
|
|
|
|
|
8
|
|
|
import numpy as np |
|
9
|
|
|
import pandas as pd |
|
10
|
|
|
|
|
11
|
|
|
|
|
12
|
|
|
class Results: |
|
13
|
|
|
"""Results class.""" |
|
14
|
|
|
|
|
15
|
|
|
def __init__(self, results_list, opt_pros): |
|
16
|
|
|
self.results_list = results_list |
|
17
|
|
|
self.opt_pros = opt_pros |
|
18
|
|
|
|
|
19
|
|
|
self.objFunc2results = {} |
|
20
|
|
|
self.search_id2results = {} |
|
21
|
|
|
|
|
22
|
|
|
def _sort_results_objFunc(self, objective_function): |
|
23
|
|
|
best_score = -np.inf |
|
24
|
|
|
best_para = None |
|
25
|
|
|
search_data = None |
|
26
|
|
|
|
|
27
|
|
|
search_data_list = [] |
|
28
|
|
|
|
|
29
|
|
|
for results_ in self.results_list: |
|
30
|
|
|
nth_process = results_["nth_process"] |
|
31
|
|
|
|
|
32
|
|
|
opt = self.opt_pros[nth_process] |
|
33
|
|
|
objective_function_ = opt.objective_function |
|
34
|
|
|
search_space_ = opt.s_space() |
|
35
|
|
|
params = list(search_space_.keys()) |
|
36
|
|
|
|
|
37
|
|
|
if objective_function_ != objective_function: |
|
38
|
|
|
continue |
|
39
|
|
|
|
|
40
|
|
|
if results_["best_score"] > best_score: |
|
41
|
|
|
best_score = results_["best_score"] |
|
42
|
|
|
best_para = results_["best_para"] |
|
43
|
|
|
|
|
44
|
|
|
search_data = results_["search_data"] |
|
45
|
|
|
search_data["eval_times"] = results_["eval_times"] |
|
46
|
|
|
search_data["iter_times"] = results_["iter_times"] |
|
47
|
|
|
|
|
48
|
|
|
search_data_list.append(search_data) |
|
49
|
|
|
|
|
50
|
|
|
if len(search_data_list) > 0: |
|
51
|
|
|
search_data = pd.concat(search_data_list) |
|
52
|
|
|
|
|
53
|
|
|
self.objFunc2results[objective_function] = { |
|
54
|
|
|
"best_para": best_para, |
|
55
|
|
|
"best_score": best_score, |
|
56
|
|
|
"search_data": search_data, |
|
57
|
|
|
"params": params, |
|
|
|
|
|
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
def _get_result(self, id_, result_name): |
|
61
|
|
|
if id_ not in self.objFunc2results: |
|
62
|
|
|
self._sort_results_objFunc(id_) |
|
63
|
|
|
|
|
64
|
|
|
search_data = self.objFunc2results[id_][result_name] |
|
65
|
|
|
|
|
66
|
|
|
return search_data |
|
67
|
|
|
|
|
68
|
|
|
def best_para(self, id_): |
|
69
|
|
|
"""Best Para function.""" |
|
70
|
|
|
best_para_ = self._get_result(id_, "best_para") |
|
71
|
|
|
|
|
72
|
|
|
if best_para_ is not None: |
|
73
|
|
|
return best_para_ |
|
74
|
|
|
|
|
75
|
|
|
raise ValueError("objective function name not recognized") |
|
76
|
|
|
|
|
77
|
|
|
def best_score(self, id_): |
|
78
|
|
|
"""Best Score function.""" |
|
79
|
|
|
best_score_ = self._get_result(id_, "best_score") |
|
80
|
|
|
|
|
81
|
|
|
if best_score_ != -np.inf: |
|
82
|
|
|
return best_score_ |
|
83
|
|
|
|
|
84
|
|
|
raise ValueError("objective function name not recognized") |
|
85
|
|
|
|
|
86
|
|
|
def search_data(self, id_): |
|
87
|
|
|
"""Search Data function.""" |
|
88
|
|
|
search_data = self._get_result(id_, "search_data") |
|
89
|
|
|
|
|
90
|
|
|
if search_data is not None: |
|
91
|
|
|
return search_data |
|
92
|
|
|
|
|
93
|
|
|
raise ValueError("objective function name not recognized") |
|
94
|
|
|
|
SimonBlanke /
Hyperactive
Loading history...