-
Notifications
You must be signed in to change notification settings - Fork 0
/
exp_analytics.py
99 lines (76 loc) · 4.21 KB
/
exp_analytics.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
import json
import os
import numpy as np
from scratch.strategic import CEKDLossPlugin, FullyConnectedNetwork
import matplotlib.pyplot as plt
def get_best(scenario, strat, criteria, *params_to_compare):
folder = "../Resultados/" + scenario + '/' + strat
exp_dict = dict()
exp_grouped_dict = dict()
for i in os.listdir(folder):
with open(os.path.join(folder, i, 'metadata.json')) as new_file:
file_contents = new_file.read()
exp_dict[i] = json.loads(file_contents)
for i in exp_dict.values():
#dict lr, decay
key = tuple([i['training_parameters'][j] for j in params_to_compare])
if key not in exp_grouped_dict.keys():
exp_grouped_dict[key] = [i['results'][criteria][-1]]
else:
exp_grouped_dict[key] += [i['results'][criteria][-1]]
for j in exp_grouped_dict:
exp_grouped_dict[j] = np.mean(exp_grouped_dict[j])
print("Sequência de parâmetros: ", [j for j in params_to_compare])
print("Melhor grupo de parâmetros é:",sorted(exp_grouped_dict.items(), key=lambda x: x[1])[-1])
return {(scenario, strat): sorted(exp_grouped_dict.items(), key=lambda x: x[1])[-1]}
def retrieve_values(scenario, strat, criteria, param_values):
folder = "../Resultados/" + scenario + '/' + strat
exp_dict = dict()
exp_grouped_dict = dict()
if len(param_values) == 2:
params_to_compare = ['learning_rate', 'weight_decay']
else:
params_to_compare = ['learning_rate', 'weight_decay', 'plasticity_factor']
for i in os.listdir(folder):
with open(os.path.join(folder, i, 'metadata.json')) as new_file:
file_contents = new_file.read()
exp_dict[i] = json.loads(file_contents)
for i in exp_dict.values():
#dict lr, decay
key = tuple([i['training_parameters'][j] for j in params_to_compare])
if key not in exp_grouped_dict.keys():
exp_grouped_dict[key] = [i['results'][criteria]]
else:
exp_grouped_dict[key] += [i['results'][criteria]]
return exp_grouped_dict[param_values]
if __name__ == '__main__':
best_param_dict = dict()
for scenario in ["DSADS_TI", "HAPT_TI", "UCIHAR_TI", "PAMAP_TI"]:
for strat in ["naive", 'waadb', 'wamdf']:
print(f"Results below for: {scenario, strat}")
best_param_dict.update(get_best(scenario, strat, "F1-macro de teste por tarefa", 'learning_rate', 'weight_decay'))
for scenario in ["DSADS_TI", "HAPT_TI", "UCIHAR_TI", "PAMAP_TI"]:
for strat in ['waadb_plasticity', 'wamdf_plasticity']:
print(f"Results below for: {scenario, strat}")
best_param_dict.update(get_best(scenario, strat, "F1-macro de teste por tarefa", 'learning_rate', 'weight_decay', 'plasticity_factor'))
bottom_limits = {"DSADS_TI": 0.02, "HAPT_TI": 0.023, "UCIHAR_TI": 0.225, "PAMAP_TI": 0.0449}
upper_limits = {"DSADS_TI": 0.7787, "HAPT_TI": 0.7084, "UCIHAR_TI": 0.8894, "PAMAP_TI": 0.8210}
for scenario in ["DSADS_TI", "HAPT_TI", "UCIHAR_TI", "PAMAP_TI"]:
best_params = [(i[1], best_param_dict[i]) for i in best_param_dict if scenario in i]
plt.figure(figsize = (12, 12))
xsize = 10
for j in best_params:
if "naive" not in j[0]:
all_matches = np.array(retrieve_values(scenario, j[0], "F1-macro de teste por tarefa", j[1][0]))
mean = np.mean(all_matches, axis = 0)
std = np.std(all_matches, axis = 0)
xsize = mean.shape[0]
plt.errorbar([i for i in range(mean.shape[0])], list(mean), list(std), label = j[0], marker = '*')
plt.plot([i for i in range(xsize)], [bottom_limits[scenario] for i in range(xsize)], label = "Bottom limit")
plt.plot([i for i in range(xsize)], [upper_limits[scenario] for i in range(xsize)], label = "Upper limit")
plt.ylim(0, 1)
plt.xlim(0 - 0.3, xsize - 0.7)
plt.xlabel("Experiência", fontsize = 20)
plt.ylabel("F1-Macro de teste", fontsize = 20)
plt.xticks([i for i in range(xsize)], fontsize = '15')
plt.yticks(np.arange(0, 1, step = 0.1), fontsize = '15')