-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathMain_sparse.py
314 lines (258 loc) · 13 KB
/
Main_sparse.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
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
# -*- coding: utf-8 -*-
"""
Created on Fri Mar 18 16:27:22 2022
@author: mahom
"""
import math
import sys
import numpy as np
import numpy.matlib
import time
import scipy as SP
import os
import torch
import gpytorch
from matplotlib import pyplot as plt
import pathlib as Path
from os import listdir
import pandas as pd
ProjectPath = Path.Path.cwd()
utilsPath = Path.Path.joinpath(ProjectPath,"utils")
probsUtilsPath = Path.Path.joinpath(ProjectPath,"Prob-utils")
ResultsPath = Path.Path.joinpath(ProjectPath,"Results")
UTIL_DIR = utilsPath
sys.path.append(
str(UTIL_DIR)
)
UTIL_DIR_GEN = probsUtilsPath
sys.path.append(
str(UTIL_DIR_GEN)
)
RESULTS_DIR_GEN = ResultsPath
sys.path.append(
str(RESULTS_DIR_GEN)
)
from EvaluateConfidenceIntervals import EvaluateConfidenceIntervals
from StandarizeData import StandarizeData
from DeStandarizeData import DeStandarizeData
from MAPE import MAPE
from GP24I_v4 import GP24I
from GPind import GPind
from GPind_ori import GPind_ori
from predGPind_ori import predGPind_ori
from GPind_lap import GPind_lap
from predGPK import predGPK
from predGPind_lap import predGPind_lap
from GPKtorch import GPKtorch
from GPK_sparse_torch import GPK_sparse_torch
from predGPind import predGPind
from load_obj import load_obj
from save_obj import save_obj
from sklearn.metrics import r2_score
from data_to_torch import data_to_torch
from norm2laplace import norm2laplace
from EvaluateConfidenceIntervals_Laplace import EvaluateConfidenceIntervals_Laplace
from outliers_removal import outliers_removal
from load_configuration import load_configuration
from load_configuration_job_array import load_configuration_job_array
from print_configuration import print_configuration
from correcting_factor_cov import correcting_factor_cov
from correcting_factor_cov_gamma import correcting_factor_cov_gamma
from predictive_variance_white import predictive_variance_white
from print_extra_methods import print_extra_methods
from GP24I_v4 import GP24I
from to_torch import to_torch
from predGPK_sp import predGPK_sp
from print_results_ic import print_results_ic
from PlotDayDecomposition_sparse import PlotDayDecomposition_sparse
# #Load Power Load Data =========================================================
# #==============================================================================
method = "NMF" # Full
methodfile = 'NMF'
kernel_type = "rbf"
forecast_method = "gpk_sp" # gp_ind_ori/gp_ind/gpk/gp_ind_laplace/gpmt/gpk_sp
option_lv = "gp_ind_ori" # gp_ind_ori/gpmt
if forecast_method == "gpk":
name_forecast_method = forecast_method +"_" +option_lv
else:
name_forecast_method = forecast_method
EXPERIMENT = 2 # This has to do with the verion of the NMF generated
TaskNumber = 24
Stand = False
#folder_data_name = "Exp_"+str(EXPERIMENT)
#folder_data_name = "BuenosResNMF"
folder_data_name = "Exp_interp_1_" + str(EXPERIMENT)
#LOCATIONS = ['ME','CT','NH','RI','NEMASSBOST','SEMASS','VT','WCMASS']
datapath = Path.Path.joinpath(ProjectPath,"Data",folder_data_name,'NMF')
DATAPATH = str(datapath)
onlyfilesALL = [f for f in listdir(DATAPATH) if f.endswith('.pkl')]
[onlyfiles,opt_parameters,forecast_method] = load_configuration(sys.argv,onlyfilesALL,forecast_method)
#[onlyfiles,opt_parameters, forecast_method, option_lv] = load_configuration_job_array(sys.argv,onlyfilesALL)
gpytorch.settings.max_cg_iterations._set_value(10000)
RESULTS = {}
for archivo in range(len(onlyfiles)):
Results = {'R224': [],'mapes':[],'mapemedio':[],'training_time':[],'test_time':[],
'Ypred':[],'Vpred':[],'likelihood':[],'ICs':[],'ICs_lap1':[],'ICs_lap2':[],'gpk':[]}
# LOAD DATA================================================================
file_name = onlyfiles[archivo]
file_path = Path.Path.joinpath(datapath,file_name)
FILE_PATH = str(file_path)
DATA = load_obj(FILE_PATH)
DATA = data_to_torch(DATA)
print(FILE_PATH)
if 'metaNMFsparse' in DATA:
metaNMFsparse = DATA['metaNMFsparse']
metaNMFsparse_test = DATA['metaNMFsparse_test']
XTrain = DATA['X_Train_Val'].T # N x F ### torch.from_numpy
YTrain = DATA['Y_Train_Val']
XTest = DATA['X_Test'].T # N x F
#YTest = DATA['Y_Test'] # N x K
Ntest = XTest.size(0)
n_task = YTrain.size(1)
YTest = torch.ones(Ntest,n_task)
YTest_24 = DATA['Y_Test_24'] # N x T
YTrain_24 = DATA['Y_Train_Val_24']
TaskNumber = np.size(DATA['Wtrain_load'],1)
WTrain = to_torch(DATA['Wtrain_load'])
Stds_train_load = DATA['Stds_train_load']
Ntest = np.size(YTest_24,0)
Ntrain = np.size(YTrain_24,0)
PlotDayDecomposition_sparse(YTrain_24,WTrain,YTrain,Stds_train_load)
#[XTrain,XTest,YTrain_24,YTest_24] = outliers_removal(XTrain,XTest,YTrain_24,YTest_24)
# nn = 100
# YTrain_24_std = np.divide(YTrain_24,np.matlib.repmat(Stds_train_load.T,Ntrain,1))
# YTrain24M = YTrain_24[0:nn,:]
# YTrainstd24M = YTrain_24_std[0:nn,:]
# XTrainM = XTrain[0:nn,:]
# YTrainM = YTrain[0:nn,:]
# XTrain = XTrainM
# YTrain = YTrainM
# YTrain_24 = YTrain24M
# YTrain_24_std = YTrainstd24M
# NORMATLIZATION================================================================
if forecast_method == "gpk_sp":
[XTrain_S, YTrain_K_S , XTest_S, YTest_K_S,scalerX, scalerY_K]=StandarizeData(XTrain,YTrain, XTest,YTest,Standarize = Stand)
else:
[XTrain_S, YTrain_24_S , XTest_S, YTest_24_S,scalerX, scalerY_24]=StandarizeData(XTrain,YTrain_24, XTest,YTest_24,Standarize = Stand)
start = time.time()
# TRAINING================================================================
#==========================================================================
if forecast_method == "gp_ind_ori":
[M,L,RESULTS,model,like] = GPind_ori(XTrain_S,YTrain_24_S,24,kernel_type,opt_parameters)
#elif forecast_method == "gpk":
end = time.time()
training_time = end-start
#=========================================================================
if forecast_method == "gpk":
K = YTrain.size(1)
[M,L,RESULTS,model,like,ind_val] = GPKtorch(XTrain_S,YTrain_K_S,WTrain,K,kernel_type,option_lv,opt_parameters)
if forecast_method == "gpk_sp":
K = YTrain.size(1)
Ntest = XTest.size(0)
[OptModel,OptLikelihood, RESULTS,IDRegressorTypes, IDRegressorTypes_test,X_test,Labels,Labels_test,Indices_test,Ind_Val] = GPK_sparse_torch(XTrain_S,YTrain_K_S,XTest_S,WTrain,K,kernel_type,option_lv,opt_parameters,metaNMFsparse,metaNMFsparse_test)
end = time.time()
training_time = end-start
# TESTING==================================================================
#==========================================================================
start = time.time()
if forecast_method == "gpk_sp":
[YPredictedS_KgpS,VPredicted_Kgp_S] = predGPK_sp(Ntest,K,OptModel,OptLikelihood, IDRegressorTypes, IDRegressorTypes_test,X_test,Labels,Labels_test,Indices_test)
[_, YPredicted_24gp_K,VPredicted_24gp_K]=DeStandarizeData(YTest_K_S,YPredictedS_KgpS,scalerY_K,VPredicted_Kgp_S,Standarize = Stand)
end = time.time()
testing_time = end-start
#==============================================================================
#==============================================================================
#==============================================================================
print_configuration(file_name,name_forecast_method,kernel_type,EXPERIMENT,Stand,folder_data_name)
if forecast_method == "gpk":
# TRANSFORMATION====
S2norm = torch.pow(Stds_train_load,2)
Snorm = Stds_train_load.T.repeat(Ntest,1)
Snorm_tr = Stds_train_load.T.repeat(Ntrain,1)
#ErrorValidation_std_P = torch.stack(RESULTS['ValidationPredictiveErrors'],dim =1)
YPredicted_24gp = (YPredicted_24gp_K@WTrain.T)*Snorm
VPredicted_24gp = torch.zeros((Ntest,24))
# if 'ValidationErrors' in RESULTS:
# ErrorValidation_std = torch.stack(RESULTS['ValidationErrors'],dim =1)
# Nval = ErrorValidation_std.size(0)
# Snorm_val = Stds_train_load.T.repeat(Nval,1)
# NoiseEstimation_Variance3 = torch.var((ErrorValidation_std@WTrain.T)*Snorm_val,axis=0)
# ind_a = np.random.permutation(range(0,Ntrain))[0:100]
# a = correcting_factor_cov(model,WTrain,YTrain_24[ind_a,:],XTrain_S[ind_a,:],option_lv,scalerY_K,NoiseEstimation_Variance3,Stds_train_load )
# a_gamma = correcting_factor_cov_gamma(model,WTrain,YTrain_24[ind_a,:],XTrain_S[ind_a,:],option_lv,scalerY_K,NoiseEstimation_Variance3,Stds_train_load )
# for ss in range(0,Ntest):
# VPredicted_24gp[ss,:] = (torch.diag(WTrain@torch.diag(VPredicted_24gp_K[ss,:])@WTrain.T)*(S2norm.ravel()) + NoiseEstimation_Variance3)*a
# VPredicted_24gp_white = predictive_variance_white(VPredicted_24gp_K,WTrain,NoiseEstimation_Variance3,S2norm)
print_extra_methods(Stds_train_load,Ntest,Ntrain,WTrain,YTrain_24,YTest_24,XTrain_S,YPredicted_24gp_K,VPredicted_24gp_K,option_lv,scalerY_K,RESULTS,model,DATA)
elif forecast_method == "gp_ind_ori":
[_, YPredicted_24gp,VPredicted_24gp]=DeStandarizeData(YTest_24_S,YPredicted_24gp_S,scalerY_24,VPredicted_24gp_S,Standarize = Stand)
[ICs,ICs_lap1,ICs_lap2] = print_results_ic(YPredicted_24gp,YTest_24,VPredicted_24gp,"gp_ind_ori")
elif forecast_method == "gpk_sp":
S2norm = torch.pow(Stds_train_load,2)
Snorm = Stds_train_load.T.repeat(Ntest,1)
Snorm_tr = Stds_train_load.T.repeat(Ntrain,1)
#ErrorValidation_std_P = torch.stack(RESULTS['ValidationPredictiveErrors'],dim =1)
YPredicted_24gp = (YPredicted_24gp_K@WTrain.T)*Snorm
VPredicted_24gp = torch.zeros((Ntest,24))
if Stand == True:
stdgp = torch.sqrt(torch.tensor(scalerY_K.var_)).float()
else:
stdgp = torch.sqrt(torch.tensor(1.0)).float()
# GP predictive validation variance error============================================
Labels_val = {}
if 'ValidationPredictiveErrors' in RESULTS:
for t in range(0,len(metaNMFsparse['LabelClass'])):
Labels_val_k = metaNMFsparse['LabelClass']['task{}'.format(t+1)]
Labels_val['task{}'.format(t+1)] = Labels_val_k
E = construct_sparse_pred_error(Labels_val,Y,Ind_Val)
Var_EvalPred_std = torch.tensor([torch.var(x) for x in EvalPred])
NoiseEstimation_Variance_GPpred = (WTrain@(Var_EvalPred_std*torch.pow(stdgp,2))@WTrain.T)*torch.pow(Stds_train_load.T,2)
# GP validation variance error============================================
if 'ValidationErrors' in RESULTS:
ErrorValidation_std = torch.stack(RESULTS['ValidationErrors'],dim =1)
ErrorValidation_aux = ErrorValidation_std*stdgp
NoiseEstimation_Variance_GPval = torch.var((ErrorValidation_aux@WTrain.T)*Snorm_val,axis=0)
for ss in range(0,Ntest):
VPredicted_24gp[ss,:] = (torch.diag(WTrain@torch.diag(VPredicted_24gp_K[ss,:])@WTrain.T)*(S2norm.ravel()) + NoiseEstimation_Variance_GPpred)
# METRICS==================================================================
mapes= MAPE(YTest_24,YPredicted_24gp)
mapemedio = torch.mean(mapes)
NTest = np.size(YTest_24,0)
R2_all = np.zeros((NTest,1))
for samp in range(0,NTest):
R2_all[samp,0] = r2_score(YTest_24[samp,:],YPredicted_24gp[samp,:])
r2_24gp = np.mean(R2_all)
# PRINT===================================================================
print('Mape Medio 24GPs indep ', mapemedio )
print('R2 24GPs i: ',r2_24gp)
if 'ValidationErrors' in RESULTS:
Lval = RESULTS['ValidationErrors']
Lval_tasks = [torch.mean(x) for x in Lval]
Lval_mean = torch.mean(torch.tensor(Lval_tasks))
print('Mean validation loss ',Lval_mean)
print('Training time: ', training_time )
print('Test time: ', testing_time)
#==========================================================================
Results['R224'] = r2_24gp
Results['mapes'] = mapes
Results['mapemedio'] = mapemedio
Results['training_time'] = training_time
Results['test_time'] = testing_time
Results['Ypred'] = YPredicted_24gp
Results['Vpred'] = VPredicted_24gp
Results['likelihood'] = like
#Results['ICs'] = ICs
#Results['ICs_lap1'] = ICs_lap1
#Results['ICs_lap2'] = ICs_lap2
if forecast_method == "gpk":
Results['Wtrain'] = WTrain
RESULTS[archivo] = Results
file_name = name_forecast_method+"_Stand_"+str(Stand)
if 'INFO' in locals():
file_name = file_name="Exp_"+str(EXPERIMENT)
file_results = Path.Path.joinpath(ResultsPath,file_name+"_results")
file_model = Path.Path.joinpath(ResultsPath,file_name+"_model")
file_data = Path.Path.joinpath(ResultsPath,file_name+"_data")
save_obj(RESULTS, file_results.as_posix())
save_obj(model, file_model.as_posix())
save_obj(DATA, file_data.as_posix())