|
1
|
|
|
# Author: Simon Blanke |
|
2
|
|
|
# Email: [email protected] |
|
3
|
|
|
# License: MIT License |
|
4
|
|
|
|
|
5
|
|
|
import numpy as np |
|
6
|
|
|
import pandas as pd |
|
7
|
|
|
|
|
8
|
|
|
|
|
9
|
|
|
def intersection(lst1, lst2): |
|
10
|
|
|
inter = [value for value in lst1 if value in lst2] |
|
11
|
|
|
return inter |
|
12
|
|
|
|
|
13
|
|
|
|
|
14
|
|
|
def convert_dataframe(dataframe1, search_space1, search_space2): |
|
15
|
|
|
dataframe2 = dataframe1.copy() |
|
16
|
|
|
|
|
17
|
|
|
for para1 in search_space1: |
|
18
|
|
|
if para1 not in search_space2: |
|
19
|
|
|
dataframe2.drop(para1, axis=1, inplace=True) |
|
20
|
|
|
continue |
|
21
|
|
|
|
|
22
|
|
|
search_elements1 = search_space1[para1] |
|
23
|
|
|
search_elements2 = search_space2[para1] |
|
24
|
|
|
|
|
25
|
|
|
inter = intersection(search_elements1, search_elements2) |
|
26
|
|
|
|
|
27
|
|
|
indices_A = [search_elements1.index(x) for x in inter] |
|
28
|
|
|
indices_B = [search_elements2.index(x) for x in inter] |
|
29
|
|
|
|
|
30
|
|
|
conv_dict = dict(zip(indices_A, indices_B)) |
|
31
|
|
|
|
|
32
|
|
|
col = dataframe2[para1] |
|
33
|
|
|
col_conv = col.map(conv_dict) |
|
34
|
|
|
col_conv = col_conv.dropna(how="any") |
|
35
|
|
|
col_conv = col_conv.astype(int) |
|
36
|
|
|
|
|
37
|
|
|
dataframe2[para1] = col_conv |
|
38
|
|
|
|
|
39
|
|
|
dataframe2 = dataframe2.dropna(how="any", axis=0) |
|
40
|
|
|
|
|
41
|
|
|
return dataframe2 |
|
42
|
|
|
|
|
43
|
|
|
|
|
44
|
|
|
def memory_dict2dataframe(memory_dict, search_space): |
|
45
|
|
|
columns = list(search_space.keys()) |
|
46
|
|
|
|
|
47
|
|
|
if not bool(memory_dict): |
|
48
|
|
|
print("Memory dictionary is empty.") |
|
49
|
|
|
return pd.DataFrame([], columns=columns) |
|
50
|
|
|
|
|
51
|
|
|
pos_tuple_list = list(memory_dict.keys()) |
|
52
|
|
|
result_list = list(memory_dict.values()) |
|
53
|
|
|
|
|
54
|
|
|
results_df = pd.DataFrame(result_list) |
|
55
|
|
|
np_pos = np.array(pos_tuple_list) |
|
56
|
|
|
|
|
57
|
|
|
pd_pos = pd.DataFrame(np_pos, columns=columns) |
|
58
|
|
|
dataframe = pd.concat([pd_pos, results_df], axis=1) |
|
59
|
|
|
|
|
60
|
|
|
return dataframe |
|
61
|
|
|
|
|
62
|
|
|
|
|
63
|
|
|
def dataframe2memory_dict(dataframe, search_space): |
|
64
|
|
|
columns = list(search_space.keys()) |
|
65
|
|
|
|
|
66
|
|
|
if dataframe.empty: |
|
67
|
|
|
print("Memory dataframe is empty.") |
|
68
|
|
|
return {} |
|
69
|
|
|
|
|
70
|
|
|
positions = dataframe[columns] |
|
71
|
|
|
scores = dataframe.drop(columns, axis=1) |
|
72
|
|
|
|
|
73
|
|
|
scores = scores.to_dict("records") |
|
74
|
|
|
positions_list = positions.values.tolist() |
|
75
|
|
|
|
|
76
|
|
|
# list of lists into list of tuples |
|
77
|
|
|
pos_tuple_list = list(map(tuple, positions_list)) |
|
78
|
|
|
memory_dict = dict(zip(pos_tuple_list, scores)) |
|
79
|
|
|
|
|
80
|
|
|
return memory_dict |
|
81
|
|
|
|
|
82
|
|
|
|
SimonBlanke /
Optimization-Metadata
