-
Notifications
You must be signed in to change notification settings - Fork 0
/
065_POST_PROCESSING_ENERGY.py
166 lines (126 loc) · 6.02 KB
/
065_POST_PROCESSING_ENERGY.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
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
exec(open("Utils.py").read(), globals())
# exec(open("061_POST_PROCESSING.py").read(), globals())
import random
'''CARICAMENTO DATI '''
data = pd.read_csv( 'results/MODELING/CLASSIFICATION/metrics.csv')
data_NN = pd.read_csv( 'results/MODELING/CLASSIFICATION/NEURAL_NETWORK/metrics.csv')
data = pd.concat( [ data, data_NN])
#data = data_NN.copy()
############## DELETING ################
data.shape
data = data[(data.Method != 'DECISION_TREE')]
data.shape
########################################
dir_images = 'Images/'
create_dir(dir_images)
dir_dest = dir_images + 'MODELING/'
create_dir( dir_dest )
''' SELEZIONE DEL MIGLIOR MODELLO '''
##################################################
################# BEST MODEL #####################
##################################################
best_results_ACC = pd.DataFrame( columns = data.columns )
for model in data.Model.unique().tolist():
#model = data.Model.unique().tolist()[3]
current_data = data[ data.Model == model ]
maximum = max(current_data.Accuracy)
ix_max = current_data.Accuracy.nlargest(1).index
row = current_data.ix[ix_max]
if model not in best_results_ACC.Model.tolist():
best_results_ACC = best_results_ACC.append( row, ignore_index = True )
best_results_ACC = best_results_ACC.round(decimals = 4)
#best_results_ACC.to_csv( dir_dest + 'best_results_ACC.csv', index = False)
best_results_AUC = pd.DataFrame( columns = data.columns )
for model in data.Model.unique().tolist():
current_data = data[ data.Model == model ]
maximum = max(current_data.AUC)
ix_max = current_data.AUC.nlargest(1).index
row = current_data.ix[ix_max]
best_results_AUC = best_results_AUC.append( row, ignore_index = True )
best_results_AUC = best_results_AUC.round(decimals = 4)
# best_results_AUC.to_csv( dir_dest + 'best_results_AUC.csv', index = False)
''' ENERGY ANALYSIS'''
#########################################################
################# ENERGY - ACCURACY #####################
#########################################################
data = pd.read_csv( 'results/MODELING/CLASSIFICATION/subset_metrics.csv')
data_NN = pd.read_csv( 'results/MODELING/CLASSIFICATION/NEURAL_NETWORK/subset_metrics.csv')
data = pd.concat( [ data, data_NN])
dir_dest = dir_images + 'MODELING/'
dir_dest = dir_dest + 'ENERGY/'
create_dir( dir_dest )
data.shape
data.Method.unique()
data.Model.unique()
print data.columns
data = data[ data.Treshold == 0.5 ]
#data['Model_(treshold)'] = data.Model + ' (' + data['Treshold'].astype(str) + ')'
colors = ['blue', 'hotpink', 'navy', 'beige', 'lavender' ,'k', 'gold', 'green', 'aqua']
for i in range( len(best_results_ACC)):
color = colors[ i ]
model = best_results_ACC.ix[i, :].Model
if model == 'KNN':
method = 'GBM'
n_var = 36
else:
method = best_results_ACC.ix[i, :].Method
n_var = best_results_ACC.ix[i, :].n_variables
#accuracy = best_results.ix[i, :].ACCURACY
current_data = data[ (data.Model == model) & (data.Method == method) & (data.n_variables == n_var)]
#current_data = current_data[ current_data.Energy < 10000]
current_data = current_data.sort_values( by = 'Energy')
#plt.xticks( np.log2(current_data.Energy), 'log' + current_data.Energy)
#corr = np.around( np.corrcoef( np.log2(current_data.Energy), current_data.Accuracy )[0,1], 2)
print current_data.Model.unique(), np.around( scipy.stats.pearsonr( np.log2(current_data.Energy),
current_data.Accuracy ), 4)
############
plt.plot( np.log2(current_data.Energy), current_data.Accuracy, 'bs-', color = color, label = model)
plt.title('Accuracy and energy classes')
plt.ylabel('Accuracy')
plt.xlabel('Log2(energy)')
plt.legend()
plt.savefig(dir_dest + '051_LOG_Energy_performance.png')
plt.close()
for i in range( len(best_results_ACC)):
color = colors[ i ]
model = best_results_ACC.ix[i, :].Model
method = best_results_ACC.ix[i, :].Method
n_var = best_results_ACC.ix[i, :].n_variables
#accuracy = best_results.ix[i, :].ACCURACY
current_data = data[ (data.Model == model) & (data.Method == method) & (data.n_variables == n_var)]
#current_data = current_data[ current_data.Energy < 10000]
current_data = current_data.sort_values( by = 'Energy')
plt.plot( current_data.Energy, current_data.Accuracy, 'bs-', color = color, label = model)
#plt.xticks( np.log2(current_data.Energy), 'log' + current_data.Energy)
plt.title('Performance modelli per diversi livelli di energia')
plt.ylabel('Accuratezza')
plt.xlabel('Energy (MEV)')
plt.legend()
plt.savefig(dir_dest + '052_Energy_performance.png')
plt.close()
# for energy in data.Energy.unique().tolist():
# # energy = data.Energy.unique().tolist()
# current_dir = dir_dest + str(energy) + '/'
# create_dir(current_dir)
# energy_data = data[ data.Energy == energy ]
# for method in data.Method.unique().tolist():
# # method = data.Method.unique().tolist()[0]
# current_data = energy_data[energy_data.Method == method]
# models = current_data.Model.unique().tolist()
# colors = ['blue', 'hotpink', 'navy', 'orange', 'k', 'gold', 'green', 'aqua']
# i = 0
# for model in current_data.Model.unique().tolist():
# # model = data.Model.unique().tolist()[2]
# current_data_model = current_data[current_data.Model == model]
# # print model, method, nvars
# # color = random.choice( colors )
# # colors.remove( color )
# plt.plot(current_data_model.n_variables, current_data_model.Accuracy, 'bs-', color=colors[i], label=model)
# plt.title('Energy: ' + str(energy) + ' MEV ' + method)
# # print i
# i = i + 1
# plt.style.use('seaborn-darkgrid')
# plt.ylabel('Accuratezza')
# plt.legend(loc='center left', bbox_to_anchor=(1.0, 0.5))
# plt.savefig(current_dir + '053_' + str(energy) + '_' + method + '.png', bbox_inches="tight")
# plt.close()