from sklearn.model_selection import train_test_split
from keras.models import Sequential, Model, model_from_json
from keras.layers import Input, LSTM, Dropout, Dense
from keras.optimizers import Adam
from keras.callbacks import ModelCheckpoint
from fileRead import file_read
# np.random.seed(0)
n_step = 50
file_path = sys.argv[1]
benchmark = file_path.split('_')[0]
# --------------------read data--------------------
seqTest, targetsTest = file_read(n_step, file_path)
# --------------------load model--------------------
lstm_model = model_from_json(
open('models/lstm_model_%s.json' % benchmark).read())
lstm_model.load_weights("weights/lstm_weights_%s.hdf5" % benchmark)
# --------------------evaluate--------------------
print 'evaluating...'
seq_num = xrange(len(seqTest))
A = 0
B = 0
C = 0
D = 0
print seqTest.shape
for test_num in seq_num:
n_step = 50
n_epoch = 100
n_hidden = 256
dropout_flag = 0
learning_rate = 0.01
decay_rate = learning_rate / n_epoch
# --------------------read data--------------------
if (len(sys.argv) == 3):
# 2 arguments
file_path = sys.argv[1]
file_path2 = sys.argv[2]
benchmark = file_path.split('_')[0]
seq, targets, seqTest, targetsTest = file_read(n_step, file_path,
file_path2)
print seq.shape, targets.shape
print seqTest.shape, targetsTest.shape
n_out = seq.shape[2]
else:
# 1 argument
file_path = sys.argv[1]
benchmark = file_path.split('_')[1]
seq, targets = file_read(n_step, file_path)
seq, seqTest, targets, targetsTest = train_test_split(seq,
targets,
test_size=0.2)
print seq.shape, targets.shape
print seqTest.shape, targetsTest.shape
n_out = seq.shape[2]
# np.random.seed(0)
n_step = 50
file_path = sys.argv[1]
benchmark = file_path.split('_')[1]
pos = file_path.split('_')[2]
if file_path.split('_')[0] == 'inputdata':
faulty_label = 0
else:
faulty_label = 1
# --------------------read data--------------------
seqTest, targetsTest = file_read(n_step,
file_path,
hasLabel=False,
label=faulty_label)
if faulty_label:
# affected for inputerror
file_path2 = sys.argv[2]
affected_raw = np.loadtxt(file_path2,
dtype='int64').reshape(-1, n_step,
seqTest.shape[2])
affected = np.zeros((affected_raw.shape[0], n_step - 1, seqTest.shape[2]),
dtype='int64')
for i in range(affected.shape[0]):
affected[i] = np.delete(affected_raw[i], 0, axis=0)
# --------------------load model--------------------
lstm_model = model_from_json(
open('models/lstm_model_%s.json' % benchmark).read())