def GetPreviousMonthsLog(self):
lastMonth = self.AddMonths(datetime.now(), -1)
logFilePath = get_logfile_path(self.logFolder, lastMonth.year,
lastMonth.month)
if path.exists(logFilePath):
with open(logFilePath, 'r') as f:
data = read_file(logFilePath)
jsonData = json.loads(data)
return jsonData['results']
else:
return None
def run(opt_method, path, seed, eps, max_iter, *args):
x, y = read_file(path)
M = LogisticRegression(x, y, seed)
w_k = opt_method(M, eps, max_iter, *args)
f_val = M.calc_func(w_k)
grad = M.calc_grad(w_k)
grad_norm = LA.norm(grad)
res = {
'f_opt': f_val,
'grad_norm': grad_norm,
'rk': M.r_k[-1],
'oracle_calls': M.oracle_calls,
'solution': w_k.tolist(),
'time': M.total_time}
return res
def LogResultsToFile(self, results):
now = datetime.now()
currentLogFolder = path.join(self.logFolder, str(now.year))
logFilePath = get_logfile_path(self.logFolder, now.year, now.month)
if not path.exists(currentLogFolder):
makedirs(currentLogFolder)
if path.exists(logFilePath):
with open(logFilePath, 'r') as f:
data = read_file(logFilePath)
jsonData = json.loads(data)
jsonData['results'] += [results]
else:
jsonData = {'results': [results]}
with open(logFilePath, 'w+') as f:
f.write(json.dumps(jsonData))
def parse():
content = read_file(argv[1])
rules = []
for line in content:
if line[0] == "?":
queries = get_qf(line, "?")
elif line[0] == "=":
facts = get_qf(line, "=")
elif line != "\n":
line = re.sub('[\s]', '', line)
if "#" in line:
line = line.split("#")[0]
rules.append(line)
ctx = {
"queries": queries,
"facts": facts,
"rules": rules
}
return ctx
def predict_func(filename):
data_filename = os.path.join("data", filename + '.csv')
output_filename = os.path.join("out", filename + '.txt')
length, list_open, list_high, list_low, list_close = read_file(
data_filename)
length, input_data, label_data = normalize_data(length, list_open,
list_high, list_low,
list_close)
data_input, data_close = [], []
for j in range(length - seq_length):
_x = input_data[j:j + seq_length]
_y = label_data[j + seq_length]
data_input.append(_x)
data_close.append(_y)
train_size = int(length * 0.7) - 1
#start index 1 for use prev_data for costing
train_input, test_input = np.array(data_input[1:train_size]), np.array(
data_input[train_size:])
train_close, test_close = np.array(data_close[1:train_size]), np.array(
data_close[train_size:])
train_close_prev = np.array(data_close[0:train_size - 1])
X = tf.placeholder(tf.float32, [None, seq_length, data_dim])
Y = tf.placeholder(tf.float32, [None, 1])
Y_prev = tf.placeholder(tf.float32, [None, 1])
# build a LSTM network
cell = tf.contrib.rnn.BasicLSTMCell(num_units=hidden_dim,
forget_bias=1.0,
state_is_tuple=True)
#cell = tf.contrib.rnn.DropoutWrapper(cell, output_keep_prob = 0.5)
outputs, _states = tf.nn.dynamic_rnn(cell, X, dtype=tf.float32)
Y_pred = tf.contrib.layers.fully_connected(outputs[:, -1],
output_dim,
activation_fn=None)
cost = (Y - Y_pred) * (tf.cast(
(Y_pred - Y_prev) * (Y - Y_prev) < 0, tf.float32) * 2 + tf.cast(
(Y_pred - Y_prev) * (Y - Y_prev) >= 0, tf.float32))
# cost/loss
loss = tf.reduce_mean(tf.square(cost)) # sum of the squares
# optimizer
optimizer = tf.train.AdamOptimizer(learning_rate)
training = optimizer.minimize(loss)
# RMSE
targets = tf.placeholder(tf.float32, [None, 1])
predictions = tf.placeholder(tf.float32, [None, 1])
rmse = tf.sqrt(tf.reduce_mean(tf.square(targets - predictions)))
with tf.Session() as sess:
init = tf.global_variables_initializer()
sess.run(init)
# Training step
for k in range(iterations):
_, step_loss = sess.run([training, loss],
feed_dict={
X: train_input,
Y: train_close,
Y_prev: train_close_prev
})
# Test step
test_predict = sess.run(Y_pred, feed_dict={X: test_input})
rmse = sess.run(rmse,
feed_dict={
targets: test_close,
predictions: test_predict
})
print(test_predict[-1])
test_close = unormalize_data(test_close, list_close)
test_predict = unormalize_data(test_predict, list_close)
with open(output_filename, 'w') as f:
f.write(to_output_form(test_close, test_predict))
def __init__(self, config_file):
content = read_file(config_file)
self.op = self._get_op(content)