def first_run():
model = model_builder.build_model()
model = model_trainer.train_model(model, x_train, y_train)
result = model_tester.test_model(model, x_test, y_test)
model.save("./nn_models/mnist_nn_v1.h5")
def train_winner(winning_design, filename, testfilename, columns, targets, testtargets, train_kwargs, separator = ','):
#print("\nTraining the winning design {0}...".format(str(winning_design)))
model_file = train_model(winning_design, filename, columns, targets, comsize_third = 10, epochs = 300, separator = separator ,
**train_kwargs)
print("\nProceeding with plotting on training data...")
model_output_file = test_model(model_file, filename, targets[0], targets[1], separator, *columns)
print("\nModel output stored in {0}".format(model_output_file))
#Against the same file, just another column
#scatterplot_files(model_output_file, 0, 2, model_output_file, 1)
cmd = 'python model_tester.py "{model_file}" YOUR_TEST_FILE_HERE "{0}" "{1}" "{2}"'.format(separator, targets[0], targets[1], \
model_file = model_file)
selfcmd = ''
if testfilename is not None:
selfcmd = 'python model_tester.py "{model_file}" "{testfile}" "{0}" "{1}" "{2}"'.format(separator, testtargets[0], testtargets[1], \
model_file = model_file, testfile = testfilename)
for col in columns:
cmd += ' "{0}"'.format(col)
if testfilename is not None:
selfcmd += ' "{0}"'.format(col)
print('''
Completed.
If you wish to test your model against some test data, use this command:
{0}
To get scatter plots for test data, run this command:
{1}'''.format(cmd, selfcmd))
def train_winner(winning_design,
filename,
testfilename,
columns,
targets,
testtargets,
train_kwargs,
separator=','):
#print("\nTraining the winning design {0}...".format(str(winning_design)))
model_file = train_model(winning_design,
filename,
columns,
targets,
comsize_third=10,
epochs=300,
separator=separator,
**train_kwargs)
print("\nProceeding with plotting on training data...")
model_output_file = test_model(model_file, filename, targets[0],
targets[1], separator, *columns)
print("\nModel output stored in {0}".format(model_output_file))
#Against the same file, just another column
#scatterplot_files(model_output_file, 0, 2, model_output_file, 1)
cmd = 'python model_tester.py "{model_file}" YOUR_TEST_FILE_HERE "{0}" "{1}" "{2}"'.format(separator, targets[0], targets[1], \
model_file = model_file)
selfcmd = ''
if testfilename is not None:
selfcmd = 'python model_tester.py "{model_file}" "{testfile}" "{0}" "{1}" "{2}"'.format(separator, testtargets[0], testtargets[1], \
model_file = model_file, testfile = testfilename)
for col in columns:
cmd += ' "{0}"'.format(col)
if testfilename is not None:
selfcmd += ' "{0}"'.format(col)
print('''
Completed.
If you wish to test your model against some test data, use this command:
{0}
To get scatter plots for test data, run this command:
{1}'''.format(cmd, selfcmd))
def check_keys(my_data):
# test for various keyboard inputs
(event, background, draw_color, line_width, keep_going, screen, mat) = my_data
if event.key == pygame.K_q:
keep_going = False
elif event.key == pygame.K_c:
background.fill((255, 255, 255))
draw_pixelated(np.zeros((28, 28)), screen)
# s - for saving a digit and pattern
elif event.key == pygame.K_s:
answer = int(ask(screen, ""))
f_m = matrix_manipulate.focus_mat(mat)
shrinking_model.learn_pattern(f_m, answer, "one_zero_shrinking")
extended_model.learn_pattern(f_m, answer, "one_zero_extended")
multiplication_model.learn_pattern(f_m, answer, "one_zero_multiplication")
"""shrinking_model.learn_pattern(f_m, answer, "shrinking_model")
extended_model.learn_pattern(f_m, answer, "extended_model")
multiplication_model.learn_pattern(f_m, answer, "multiplication_model")"""
# t - for testing some real time hand write digit
elif event.key == pygame.K_t:
model_tester.test_model(my_data, 'one_zero_shrinking')
model_tester.test_model(my_data, 'one_zero_extended')
model_tester.test_model(my_data, 'one_zero_multiplication')
"""model_tester.test_model(my_data, 'multiplication_model')
model_tester.test_model(my_data, "shrinking_model")
model_tester.test_model(my_data, "extended_model")"""
background.fill((255, 255, 255))
draw_pixelated(np.zeros((28, 28)), screen)
my_data = (event, background, draw_color, line_width, keep_going)
return my_data
def main():
transformer_list = []
regex_features = True
icd9_features = False
labs_features = False
text_features = False
if regex_features:
transformer_list += [
('EF', EFTransformer('all', 1, None)),
('EF', EFTransformer('mean', 5, None)),
('EF', EFTransformer('max', 5, None)),
('LBBB', LBBBTransformer(30 * 3)),
('SR', SinusRhythmTransformer(30 * 3)),
('NYHA', NYHATransformer(30 * 3)),
('NICM', NICMTransformer(30 * 3)),
('QRS', QRSTransformer('all', 1, None)),
('QRS', QRSTransformer('mean', 5, None)),
]
if icd9_features:
transformer_list += [('Dia', ICD9_Transformer())]
if text_features:
transformer_list += [
('Car',
FeaturePipeline([('notes_transformer_car',
GetConcatenatedNotesTransformer('Car')),
('tfidf', TfidfTransformer())])),
('Lno',
FeaturePipeline([('notes_transformer_lno',
GetConcatenatedNotesTransformer('Lno')),
('tfidf', TfidfTransformer)]))
]
if labs_features:
transformer_list += [
('Enc', GetEncountersFeaturesTransformer(5)),
('Labs_Counts',
FeaturePipeline([('labs_counts_transformer',
GetLabsCountsDictTransformer()),
('dict_vectorizer', DictVectorizer())])),
('Labs_Low_Counts',
FeaturePipeline([('labs_low_counts_transformer',
GetLabsLowCountsDictTransformer()),
('dict_vectorizer', DictVectorizer())])),
('Labs_High_Counts',
FeaturePipeline([('labs_high_counts_transformer',
GetLabsHighCountsDictTransformer()),
('dict_vectorizer', DictVectorizer())])),
('Labs_Latest_Low',
FeaturePipeline([('labs_latest_low_transformer',
GetLabsLatestLowDictTransformer()),
('dict_vectorizer', DictVectorizer())])),
('Labs_Latest_High',
FeaturePipeline([('labs_latest_high_transformer',
GetLabsLatestHighDictTransformer()),
('dict_vectorizer', DictVectorizer())])),
('Labs_History',
FeaturePipeline([('labs_history_transformer',
GetLabsHistoryDictTransformer([1])),
('dict_vectorizer', DictVectorizer())]))
]
features = FeatureUnion(transformer_list)
if len(sys.argv) > 1 and unicode(sys.argv[1]).isnumeric():
data_size = min(int(sys.argv[1]), 906)
else:
data_size = 25
if len(sys.argv) > 2 and unicode(sys.argv[2]).isnumeric():
num_cv_splits = int(sys.argv[2])
else:
num_cv_splits = 5
print "Data size: " + str(data_size)
print "CV splits: " + str(num_cv_splits)
if len(sys.argv) > 3:
method = sys.argv[3]
else:
method = 'adaboost'
#method = 'lr'
#method = 'svm'
method = 'adaboost'
#method = 'cdm'
model_args = dict()
if method in ['lr', 'svm']:
if len(sys.argv) > 4 and unicode(sys.argv[4]).isnumeric():
model_args['regularization'] = float(sys.argv[4])
else:
model_args['regularization'] = 0.
if method == 'adaboost':
if len(sys.argv) > 4 and unicode(sys.argv[4]).isnumeric():
model_args['n_estimators'] = int(sys.argv[4])
else:
model_args['n_estimators'] = 50
show_progress = True
print 'Method:', method
test_model(features, data_size, num_cv_splits, method, show_progress,
model_args)
print("\nTraining a cox committee...")
comsize = 1
if len(sys.argv) > 1:
comsize = int(sys.argv[1])
model_file = train_model(filename,
columns,
targets,
separator=separator,
comsize=comsize)
print("\nProceeding with plotting on training data...")
model_output_file = test_model(model_file, filename, targets[0],
targets[1], separator, *columns)
print("\nModel output stored in {0}".format(model_output_file))
#Against the same file, just another column
#scatterplot_files(model_output_file, 0, 2, model_output_file, 1)
cmd = 'python model_tester.py "{model_file}" YOUR_TEST_FILE_HERE "{0}" "{1}" "{2}"'.format(separator, targets[0], targets[1], \
model_file = model_file)
selfcmd = ''
if testfilename is not None:
selfcmd = 'python model_tester.py "{model_file}" "{testfile}" "{0}" "{1}" "{2}"'.format(separator, testtargets[0], testtargets[1], \
model_file = model_file, testfile = testfilename)
for col in columns:
cmd += ' "{0}"'.format(col)
if testfilename is not None:
use_header=True)
print("Retrieving test data...")
unNormedTestP, T = parse_file(testdata,
inputcols=columns,
targetcols=targets,
normalize=False,
separator=',',
use_header=True)
print("Normalizing test data...")
P = normalizeArrayLike(unNormedTestP, normP)
#Scatter training data
model_output_file = test_model(model,
trainingdata,
targets[0],
targets[1],
',',
time_step_size=2,
*columns)
scatterplot_files(model_output_file, 0, 2, model_output_file, 1)
#Scatter test data
model_output_file = test_model_arrays(model,
testdata,
P,
T,
time_step_size=2)
scatterplot_files(model_output_file, 0, 2, model_output_file, 1)
def main():
transformer_list = []
if False:
transformer_list += SymptomsExtractorTransformerGenerator(
['Car', 'Lno'], 'found', None, 6 * 30).getSymptoms()
if False:
transformer_list += [('Dia', ICD9_Transformer())]
if False:
transformer_list += [
('EF', EFTransformer('all', 1, None)),
('EF', EFTransformer('mean', 5, None)),
('EF', EFTransformer('max', 5, None)),
('LBBB', LBBBTransformer()),
('SR', SinusRhythmTransformer()),
('NYHA', NYHATransformer()),
('QRS', QRSTransformer('all', 1, None)), #Bugs with QRS
]
if False:
transformer_list += [
('Car',
FeaturePipeline([('notes_transformer_car',
GetConcatenatedNotesTransformer('Car')),
('tfidf', TfidfTransformer())])),
('Lno',
FeaturePipeline([('notes_transformer_lno',
GetConcatenatedNotesTransformer('Lno')),
('tfidf', TfidfTransformer)]))
]
if False:
transformer_list += [
#('Car', FeaturePipeline([
# ('notes_transformer_car', GetLatestNotesTransformer('Car', 100)),
# ('notes_aggregator_car', DocumentConcatenatorTransformer()),
# ('trigram', CountVectorizer(ngram_range=(3,3), min_df=2))
#])),
('Car',
FeaturePipeline([('notes_transformer_car',
GetConcatenatedNotesTransformer('Car')),
('bigram',
CountVectorizer(ngram_range=(2, 2),
min_df=0.05))])),
('Lno',
FeaturePipeline([('notes_transformer_lno',
GetConcatenatedNotesTransformer('Lno')),
('bigram',
CountVectorizer(ngram_range=(2, 2),
min_df=0.05))]))
]
if True:
transformer_list += [
#('Enc', GetEncountersFeaturesTransformer(100, True)),
('Labs_Counts',
FeaturePipeline([('labs_counts_transformer',
GetLabsCountsDictTransformer()),
('dict_vectorizer', DictVectorizer())])),
('Labs_Low_Counts',
FeaturePipeline([('labs_low_counts_transformer',
GetLabsLowCountsDictTransformer()),
('dict_vectorizer', DictVectorizer())])),
('Labs_High_Counts',
FeaturePipeline([('labs_high_counts_transformer',
GetLabsHighCountsDictTransformer()),
('dict_vectorizer', DictVectorizer())])),
#('Labs_Latest_Low', FeaturePipeline([
# ('labs_latest_low_transformer', GetLabsLatestLowDictTransformer()),
# ('dict_vectorizer', DictVectorizer())
#])),
#('Labs_Latest_High',FeaturePipeline([
# ('labs_latest_high_transformer', GetLabsLatestHighDictTransformer()),
# ('dict_vectorizer', DictVectorizer())
#])),
#('Labs_History', FeaturePipeline([
# ('labs_history_transformer', GetLabsHistoryDictTransformer([1])),
# ('dict_vectorizer', DictVectorizer())
#]))
]
features = FeatureUnion(transformer_list)
if len(sys.argv) > 1 and unicode(sys.argv[1]).isnumeric():
data_size = min(int(sys.argv[1]), 906)
else:
data_size = 25
if len(sys.argv) > 2 and unicode(sys.argv[2]).isnumeric():
num_cv_splits = int(sys.argv[2])
else:
num_cv_splits = 5
print "Data size: " + str(data_size)
print "CV splits: " + str(num_cv_splits)
#method = 'lr'
#method = 'svm'
method = 'adaboost'
#method = 'cdm'
model_args = dict()
if method in ['lr', 'svm']:
if len(sys.argv) > 3 and unicode(sys.argv[3]).isnumeric():
model_args['C'] = float(sys.argv[3])
#else:
#model_args['C'] = 0.
if method == 'adaboost':
if len(sys.argv) > 3 and unicode(sys.argv[3]).isnumeric():
model_args['n_estimators'] = int(sys.argv[3])
else:
model_args['n_estimators'] = 50
show_progress = True
print 'Method:', method
test_model(features, data_size, num_cv_splits, method, show_progress,
model_args)
def main():
transformer_list = []
if False:
transformer_list += SymptomsExtractorTransformerGenerator(['Car','Lno'], 'found', None, 6*30).getSymptoms()
if False:
transformer_list += [
('Dia', ICD9_Transformer())
]
if False:
transformer_list += [
('EF', EFTransformer('all', 1, None)),
('EF', EFTransformer('mean', 5, None)),
('EF', EFTransformer('max', 5, None)),
('LBBB', LBBBTransformer()),
('SR', SinusRhythmTransformer()),
('NYHA', NYHATransformer()),
('QRS', QRSTransformer('all', 1, None)),#Bugs with QRS
]
if False:
transformer_list += [
('Car', FeaturePipeline([
('notes_transformer_car', GetConcatenatedNotesTransformer('Car')),
('tfidf', TfidfTransformer())
])),
('Lno', FeaturePipeline([
('notes_transformer_lno', GetConcatenatedNotesTransformer('Lno')),
('tfidf', TfidfTransformer)
]))
]
if False:
transformer_list += [
#('Car', FeaturePipeline([
# ('notes_transformer_car', GetLatestNotesTransformer('Car', 100)),
# ('notes_aggregator_car', DocumentConcatenatorTransformer()),
# ('trigram', CountVectorizer(ngram_range=(3,3), min_df=2))
#])),
('Car', FeaturePipeline([
('notes_transformer_car', GetConcatenatedNotesTransformer('Car')),
('bigram', CountVectorizer(ngram_range=(2,2), min_df=0.05))
])),
('Lno', FeaturePipeline([
('notes_transformer_lno', GetConcatenatedNotesTransformer('Lno')),
('bigram', CountVectorizer(ngram_range=(2,2), min_df=0.05))
]))
]
if True:
transformer_list += [
#('Enc', GetEncountersFeaturesTransformer(100, True)),
('Labs_Counts',FeaturePipeline([
('labs_counts_transformer', GetLabsCountsDictTransformer()),
('dict_vectorizer', DictVectorizer())
])),
('Labs_Low_Counts',FeaturePipeline([
('labs_low_counts_transformer', GetLabsLowCountsDictTransformer()),
('dict_vectorizer', DictVectorizer())
])),
('Labs_High_Counts', FeaturePipeline([
('labs_high_counts_transformer', GetLabsHighCountsDictTransformer()),
('dict_vectorizer', DictVectorizer())
])),
#('Labs_Latest_Low', FeaturePipeline([
# ('labs_latest_low_transformer', GetLabsLatestLowDictTransformer()),
# ('dict_vectorizer', DictVectorizer())
#])),
#('Labs_Latest_High',FeaturePipeline([
# ('labs_latest_high_transformer', GetLabsLatestHighDictTransformer()),
# ('dict_vectorizer', DictVectorizer())
#])),
#('Labs_History', FeaturePipeline([
# ('labs_history_transformer', GetLabsHistoryDictTransformer([1])),
# ('dict_vectorizer', DictVectorizer())
#]))
]
features = FeatureUnion(transformer_list)
if len(sys.argv) > 1 and unicode(sys.argv[1]).isnumeric():
data_size = min(int(sys.argv[1]), 906)
else:
data_size = 25
if len(sys.argv) > 2 and unicode(sys.argv[2]).isnumeric():
num_cv_splits = int(sys.argv[2])
else:
num_cv_splits = 5
print "Data size: " + str(data_size)
print "CV splits: " + str(num_cv_splits)
#method = 'lr'
#method = 'svm'
method = 'adaboost'
#method = 'cdm'
model_args = dict()
if method in ['lr', 'svm']:
if len(sys.argv) > 3 and unicode(sys.argv[3]).isnumeric():
model_args['C'] = float(sys.argv[3])
#else:
#model_args['C'] = 0.
if method == 'adaboost':
if len(sys.argv) > 3 and unicode(sys.argv[3]).isnumeric():
model_args['n_estimators'] = int(sys.argv[3])
else:
model_args['n_estimators'] = 50
show_progress = True
print 'Method:', method
test_model(features, data_size, num_cv_splits, method, show_progress, model_args)
def main():
features = FeatureUnion([
('Dia', icd9),
('EF', EFTransformer('all', 1, None)),
('EF', EFTransformer('mean', 5, None)),
('EF', EFTransformer('max', 5, None)),
('LBBB', LBBBTransformer()),
#('SR', SinusRhythmTransformer()),
#('Car_Doc2Vec', Doc2Vec_Note_Transformer('Car', 'doc2vec_models/car_1.model', 10, dbow_file='doc2vec_models/car_dbow.model'))
# ('QRS', QRSTransformer('all', 1, None)),#Bugs with QRS
('car_ngram',
FeaturePipeline([
('notes_car',
GetConcatenatedNotesTransformer(note_type='Car',
look_back_months=12)),
('ngram_car', CountVectorizer(ngram_range=(2, 2), min_df=.05))
]))
#('Car', FeaturePipeline([
# ('notes_transformer_car', GetConcatenatedNotesTransformer('Car')),
# ('tfidf', car_tfidf)
#])),
#('Lno', FeaturePipeline([
# ('notes_transformer_lno', GetConcatenatedNotesTransformer('Lno')),
# ('tfidf', lno_tfidf)
#])),
#('Enc', enc),
#('Labs_Counts',FeaturePipeline([
# ('labs_counts_transformer', GetLabsCountsDictTransformer()),
# ('dict_vectorizer', DictVectorizer())
#])),
#('Labs_Low_Counts',FeaturePipeline([
# ('labs_low_counts_transformer', GetLabsLowCountsDictTransformer()),
# ('dict_vectorizer', DictVectorizer())
#])),
#('Labs_High_Counts', FeaturePipeline([
# ('labs_high_counts_transformer', GetLabsHighCountsDictTransformer()),
# ('dict_vectorizer', DictVectorizer())
#])),
#('Labs_Latest_Low', FeaturePipeline([
# ('labs_latest_low_transformer', GetLabsLatestLowDictTransformer()),
# ('dict_vectorizer', DictVectorizer())
#])),
#('Labs_Latest_High',FeaturePipeline([
# ('labs_latest_high_transformer', GetLabsLatestHighDictTransformer()),
# ('dict_vectorizer', DictVectorizer())
#])),
# ('Labs_History', FeaturePipeline([
# ('labs_history_transformer', GetLabsHistoryDictTransformer([1])),
# ('dict_vectorizer', DictVectorizer())
# ])),
])
if len(sys.argv) > 1 and unicode(sys.argv[1]).isnumeric():
data_size = min(906, int(sys.argv[1]))
else:
data_size = 25
if len(sys.argv) > 2 and unicode(sys.argv[2]).isnumeric():
num_cv_splits = int(sys.argv[2])
else:
num_cv_splits = 2
method = 'lr'
#method = 'svm'
show_progress = True
test_model(features, data_size, num_cv_splits, method, show_progress)
def main():
features = FeatureUnion([
('Dia', icd9 ),
('EF', EFTransformer('all', 1, None)),
('EF', EFTransformer('mean', 5, None)),
('EF', EFTransformer('max', 5, None)),
('LBBB', LBBBTransformer()),
#('SR', SinusRhythmTransformer()),
#('Car_Doc2Vec', Doc2Vec_Note_Transformer('Car', 'doc2vec_models/car_1.model', 10, dbow_file='doc2vec_models/car_dbow.model'))
# ('QRS', QRSTransformer('all', 1, None)),#Bugs with QRS
('car_ngram', FeaturePipeline([
('notes_car', GetConcatenatedNotesTransformer(note_type='Car',look_back_months=12)),
('ngram_car', CountVectorizer(ngram_range=(2, 2), min_df=.05))
]))
#('Car', FeaturePipeline([
# ('notes_transformer_car', GetConcatenatedNotesTransformer('Car')),
# ('tfidf', car_tfidf)
#])),
#('Lno', FeaturePipeline([
# ('notes_transformer_lno', GetConcatenatedNotesTransformer('Lno')),
# ('tfidf', lno_tfidf)
#])),
#('Enc', enc),
#('Labs_Counts',FeaturePipeline([
# ('labs_counts_transformer', GetLabsCountsDictTransformer()),
# ('dict_vectorizer', DictVectorizer())
#])),
#('Labs_Low_Counts',FeaturePipeline([
# ('labs_low_counts_transformer', GetLabsLowCountsDictTransformer()),
# ('dict_vectorizer', DictVectorizer())
#])),
#('Labs_High_Counts', FeaturePipeline([
# ('labs_high_counts_transformer', GetLabsHighCountsDictTransformer()),
# ('dict_vectorizer', DictVectorizer())
#])),
#('Labs_Latest_Low', FeaturePipeline([
# ('labs_latest_low_transformer', GetLabsLatestLowDictTransformer()),
# ('dict_vectorizer', DictVectorizer())
#])),
#('Labs_Latest_High',FeaturePipeline([
# ('labs_latest_high_transformer', GetLabsLatestHighDictTransformer()),
# ('dict_vectorizer', DictVectorizer())
#])),
# ('Labs_History', FeaturePipeline([
# ('labs_history_transformer', GetLabsHistoryDictTransformer([1])),
# ('dict_vectorizer', DictVectorizer())
# ])),
])
if len(sys.argv) > 1 and unicode(sys.argv[1]).isnumeric():
data_size = min(906, int(sys.argv[1]))
else:
data_size = 25
if len(sys.argv) > 2 and unicode(sys.argv[2]).isnumeric():
num_cv_splits = int(sys.argv[2])
else:
num_cv_splits = 2
method = 'lr'
#method = 'svm'
show_progress = True
test_model(features, data_size, num_cv_splits, method, show_progress)
print("Model: {}".format(model))
#Define the data
testdata = '/home/gibson/jonask/DataSets/breast_cancer_1/n4369_targetthird.csv'
columns = ['age', 'log(1+lymfmet)', 'n_pos', 'tumsize', 'log(1+er_cyt)', 'log(1+pgr_cyt)', 'pgr_cyt_pos',
'er_cyt_pos', 'size_gt_20', 'er_cyt', 'pgr_cyt']
targets = ['time_10y', 'event_10y']
trainingdata = '/home/gibson/jonask/DataSets/breast_cancer_1/n4369_trainingtwothirds.csv'
print("Retrieving training data...")
# Normalize the test data as we normalized the training data
normP, bah = parse_file(trainingdata, inputcols = columns, targetcols = targets, normalize = False, separator = ',',
use_header = True)
print("Retrieving test data...")
unNormedTestP, T = parse_file(testdata, inputcols = columns, targetcols = targets, normalize = False, separator = ',',
use_header = True)
print("Normalizing test data...")
P = normalizeArrayLike(unNormedTestP, normP)
#Scatter training data
model_output_file = test_model(model, trainingdata, targets[0], targets[1], ',', time_step_size = 2, *columns)
scatterplot_files(model_output_file, 0, 2, model_output_file, 1)
#Scatter test data
model_output_file = test_model_arrays(model, testdata, P, T, time_step_size=2)
scatterplot_files(model_output_file, 0, 2, model_output_file, 1)
# 'BSI_A7', 'BSI_A8', 'BSI_A9', 'BSI_A10', 'BSI_A11', 'BSI_A12', 'N_A1', 'N_A2', 'N_A3',
# 'N_A4', 'N_A5', 'N_A6', 'N_A7', 'N_A8', 'N_A9', 'N_A10', 'N_A11', 'N_A12')
# targets = ['Stid', 'Event']
# testtargets = []
print("\nTraining a cox committee...")
comsize = 1
if len(sys.argv) > 1:
comsize = int(sys.argv[1])
model_file = train_model(filename, columns, targets, separator=separator, comsize=comsize)
print("\nProceeding with plotting on training data...")
model_output_file = test_model(model_file, filename, targets[0], targets[1], separator, *columns)
print("\nModel output stored in {0}".format(model_output_file))
# Against the same file, just another column
# scatterplot_files(model_output_file, 0, 2, model_output_file, 1)
cmd = 'python model_tester.py "{model_file}" YOUR_TEST_FILE_HERE "{0}" "{1}" "{2}"'.format(
separator, targets[0], targets[1], model_file=model_file
)
selfcmd = ""
if testfilename is not None:
selfcmd = 'python model_tester.py "{model_file}" "{testfile}" "{0}" "{1}" "{2}"'.format(
separator, testtargets[0], testtargets[1], model_file=model_file, testfile=testfilename
)
for col in columns:
while runs < max_runs:
episodes = 0
agent = Agent(stateCnt, actionCnt)
print("training run ", runs + 1)
while episodes < MAX_EPISODES:
env.run(agent)
episodes = episodes + 1
#ss = 0 # blah blah
if ENV_LEARN:
agent.brain.env_model.env_model.model.save("models/env_model_" +
MODEL_VER + ".h5")
agent.brain.env_model.r_model.save("models/r_model_" + MODEL_VER +
".h5")
agent.brain.controller.save('models/controller_' + MODEL_VER + ".h5")
print("testing run ", runs + 1)
d_cnt, R_cnt = test_model(VAE_VER, MODEL_VER, use_all=False, val=True)
done_counts.append(d_cnt)
R_counts.append(R_cnt)
runs += 1
#plt.plot(r_history)
#plt.show()
finally:
if ENV_LEARN:
agent.brain.env_model.env_model.model.save("models/env_model_" +
MODEL_VER + ".h5")
agent.brain.env_model.r_model.save("models/r_model_" + MODEL_VER +
".h5")
agent.brain.controller.save('models/controller_' + MODEL_VER + ".h5")
done_counts = np.asarray(done_counts)
R_counts = np.asarray(R_counts)
print("Done counts: average = ", done_counts.mean(), "max = ",